Can beehive fences help mitigate human-elephant confli

Behaviour and crop-raiding patterns of Asian Elephants

(Elephas maximus): Can beehive fences help mitigate

human-elephant conflict in Sri Lanka?

Kylie Michelle Butler

B.Sc.(Ecology and Sustainability) (Hons.)

M. Env

A thesis submitted in total fulfillment of the requirements

for the degree of Master of Philosophy (Environmental Science)

Faculty of Science

School of Environmental and Life Sciences

University of Newcastle

New South Wales, Australia

August 2019

Acknowledgement of Government Support

This research was supported by an Australian Government Research Training (TRP) Scholarship.

Statement of Originality

I hereby certify that the work embodied in the thesis is my own work, conducted under normal

supervision. The thesis contains no material which has been accepted, or is being examined, for the

award of any other degree or diploma in any university of other tertiary institution and, to the best of

my knowledge and belief, contains no material previously published or written by another person,

except where due reference has been made. I give consent to the final version of my thesis being

made available worldwide when deposited in the University’s Digital Repository, subject to the

provisions of the Copyright Act 1968 and any approved embargo.

X. Kylie M. Butler

Ethical Note

All field work was conducted under the University of Newcastle’s Animal Ethics Committee Reference

No. A-2014-413, Human Ethics Committee Reference No. H-2014-0101, with Approval from the

Department of Wildlife and Conservation, Sri Lanka, Reference No. WL/3/2/66/14.

Funding Bodies

I gratefully acknowledge funding support from the following organisations: Elephant Action League;

Save the Elephants; Chester Zoo Conservation Grant; Phoenix Zoo Conservation and Science Grant;

Rufford Small Grants for Nature Conservation; Taronga Zoo Field Conservation Grant; Zoos Victoria

Conservation Grant; and The Transpetrol Foundation.

Table of Contents

List of Figures vi

List of Tables viii

Acknowledgements x

Abstract xiii

Chapter 1: Introduction

1.1 Human-Wildlife Conflict 2

1.2 Human-elephant conflict 3

1.2.1 Human-elephant conflict in Sri Lanka 4

1.3 Characteristics and behaviour of elephants in high human-elephant conflict areas 7

1.3.1 Risk taking behaviour 7

1.4 Human-elephant conflict mitigation 9

1.4.1 Beehive fencing 11

1.5 Rationale 13

1.6 Thesis Overview 14

1.7 References 16

Chapter 2: General behaviour patterns of wild elephants utilising areas of varying anthropogenic

risk to elephants, in Sri Lanka

2.1 Abstract 29

2.2 Introduction 29

2.3 Methods 36

2.3.1 Study site 36

2.3.1.1 Wasgamuwa National Park (low risk zone) 37

2.3.1.2 Weheragala Tank (medium risk zone) 38

2.3.2 Study subjects 40

2.3.3 Data collection: elephant observations 45

2.3.4 Behaviour coding from video footage 48

2.3.5 Statistical analysis 51

2.4 Results 52

2.4.1 General behavioural activities 52

2.4.2 Specific ‘reactive’ behaviours 60

2.4.3 Influence of prior negative interactions with humans (EHI levels) 61

2.5 Discussion 64

2.5.1 General behaviour patterns in Wasgamuwa National Park and Weheragala Tank 64

2.5.2 Effects of context and sex on elephant behaviour 66

2.6 Conclusion 70

2.7 References 72

Chapter 3: Crop-raiding in a high human-elephant conflict location in Sri Lanka: an assessment of

the characteristics and patterns of crop-raiding in Dewagiriya Village

3.1 Abstract 81

3.2 Introduction 81

3.3 Methods 86

3.3.1 Study site 86

3.3.2 Data collection 89


3.4 Results 93

3.4.1 Elephant events by raid days 93

3.4.2 Factors influencing elephant presence in Dewagiriya Village 96

3.4.3 Factors influencing within-site variations in elephant events 100

3.4.4 Damage inflicted by elephants 101

3.5 Discussion 105

3.5.1 Characteristics of crop-raiding elephants 106

3.5.2 Influences of crop-raiding behaviour 107

3.5.3 Within-site variations in elephant events 110

3.6 Conclusion 112

3.7 References 113

Chapter 4: Beehive fences as a human-elephant conflict mitigation tool for farmers in rural Sri

Lanka

4.1 Abstract 119

4.2 Introduction 119

4.2.1 Human-elephant conflict mitigation 120

4.2.2 Elephant interactions with bees and beehive fences – Africa 122

4.2.3 Elephant interactions with bees and beehive fences – Asia 124

4.3 Methods 126

4.3.1 Study site 126

4.3.2 Introducing the concept of beehive fencing and participant selection for the beehive fence

trial 128

4.3.3 Establishment of beehive fences 132

4.3.4 Data collection: elephant events 136

4.3.5 Fence maintenance 137

4.3.6 Beekeeping 139

4.3.7 Cost analysis 140

4.3.8 Farmer acceptance and willingness to participate in using beehive fencing 140


4.4 Results 141

4.4.1 Elephant events and the deterrent effect of beehive fences 141

4.4.2 Fence maintenance 145

4.4.3 Beekeeping and honey harvests 147

4.4.5 Cost of beehive fences 149

4.4.6 Farmer acceptance and willingness to participate 152

4.5 Discussion 153

4.6 Conclusion 157

4.7 References 158

Chapter 5: Discussion

5.1 Introduction 167

5.2 Summary of results 167

5.3 Methodological challenges 169

5.3.1 Behavioural assessments 169

5.3.2 Profiling of crop-raiding 171

5.3.3 Beehive fence trials 171

5.4 Closing statements 173

5.5 References 175

List of Figures

Figure 1.1: Sri Lanka: National Parks and elephant distribution 5

Figure 1.2: Example of a beehive fence 12

Figure 2.1: Overview of study site 37

Figure 2.2: South-eastern section of WNP 39

Figure 2.3 Weheragala Tank ‘medium risk zone’ 39

Figure 2.4: Example from Elephant ID File. Mohan 41

Figure 2.5: Elephant with a high elephant-human interaction rating 43

Figure 2.6: Elephant with a medium elephant-human interaction rating 44

Figure 2.7: Elephant with a low elephant-human interaction rating 44

Figure 2.8: Elephant observations in Wasgamuwa National Park 46

Figure 2.9: Elephant observations to the north of Weheragala Tank 47

Figure 2.10: Elephant observations from south-west of Weheragala Tank 47

Figure 2.11: % time spent standing still (without feeding) in WNP 55

Figure 2.12: % time spent in locomotion at WT 57

Figure 2.13: % time spent feeding at WT 58

Figure 2.14: % time spent standing still (without feeding) at WT 59

Figure 2.15: Vehicles and elephants in WNP (LRZ) 67

Figure 2.16: Elephant smells towards research vehicle in WNP (LRZ) 67

Figure 2.17: Elephants fleeing from the clearing towards the forest at WT 68

Figure 3.1: Location of study site: Dewagiriya Village position in Sri Lanka 87

Figure 3.2: Map of the Dewagiriya Village study area 90

Figure 3.3: Camera trap footage of elephants in Dewagiriya Village 95

Figure 3.4: Elephant events per month based on ‘best’ model predictors 97

Figure 3.5: Elephant events per year based on ‘best’ model predictors 98

Figure 3.6: Relationship between raid day rain and elephant events based on ‘best’ model

predictors 99

Figure 3.7: Elephant damage in Dewagiriya Village 102

Figure 4.1: Diagram of a beehive fence 123

Figure 4.2: A beehive fence 123

Figure 4.3: Study site location 127

Figure 4.4: Map of Dewagiriya Village showing experimental and control households 131

Figure 4.5: Langstroth beehive secured with coconut rope 132

Figure 4.6: Langstroth beehive secured with ‘z’ clips 132

Figure 4.7: Painting fence posts with old engine oil to protect from termites 133

Figure 4.8: A beehive fence surrounding the home garden of the Jayasena family 133

Figure 4.9: Beehive fence shade roofs 134

Figure 4.10: Layout of the ten beehive fences 135

Figure 4.11: Fence maintenance issues 138

Figure 4.12: Number of elephant events: Beehive fenced (experimental) vs. non-beehive fenced

(control) home gardens 142

Figure 4.13: Frequency of elephants entering gardens during an elephant event 143

Figure 4.14: Relationship between elephant breakthroughs and number of occupied beehives 145

List of Tables

Table 2.1: Elephant-human interaction (EHI) ratings and definitions 43

Table 2.2: Elephant ethogram 48

Table 2.3: % time spent in general behavioural activities 53

Table 2.4: LMM summary: effects of context * sex on general behaviour in WNP 54

Table 2.5: Relationship between treatment * sex of elephants in the MRZ 56

Table 2.6: Relationship between sex, context and the interaction between sex and context for

reactive behaviours in WNP and WT 60

Table 2.7: Relationship between treatment * EHI levels of elephants in WNP 61

Table 2.8: Relationship between treatment * EHI levels of elephants in WNP and reactive

behaviours 62

Table 2.9: Relationship between treatment * EHI levels of elephants at WT 63

Table 2.10: Relationship between treatment * EHI levels of elephants at WT and reactive

behaviours 64

Table 3.1: Moon phase definitions 91

Table 3.2: Results from models predicting likelihood of elephant events in Dewagiriya Village

96

Table 3.3: Results from models predicting within-site variations in frequency of elephant events

per household 101

Table 3.4: Crop and plant types grown and damaged in Dewagiriya Village 103

Table 3.5: Results from models predicting likelihood of damage during elephant events in

Dewagiriya Village 104

Table 3.6: Comparison of annual crop-raiding counts per km² across HEC locations 105

Table 4.1: List of households experiencing the most negative impacts of crop-raiding in

Dewagiriya Village 130

Table 4.2: Fence establishment and data collection periods for each experimental and control

household 136

Table 4.3: Beekeeping training and workshop record 139

Table 4.4: GLMM summary: the effect of hive numbers and occupied hives on fence break-

throughs 144

Table 4.5: Presence/absence counts of different pests found in beehives 146

Table 4.6: Asian honey bee colony introductions and abscondments 148

Table 4.7: Number of active beehives per fence 148

Table 4.8: Honey harvesting records per household 149

Table 4.9: Cost of beehive fence materials and equipment per fence 150

Table 4.10: Cost estimates for colony establishment and feeding per fence 151

Table 4.11: Cost estimate for the fencing materials, tools/equipment and colonies overall (per

fence) 151

Acknowledgements

This study has been an exciting, challenging, sometimes frustrating and exceedingly beautiful journey

into the world of Sri Lankan wildlife, conservation and communities. There are many people who have

played a significant role in this adventure, and many who deserve my warmest gratitude.

My academic supervisors Dr. Andrea Griffin and Assoc. Prof. Natalie Moltchaniswkyj, I thank you for

taking me on board, for your encouragement and for pushing me beyond my academic comfort zone.

I know I haven’t been the easiest student but there are many valuable lessons I have gained from your

expertise. My industry supervisor Dr. Lucy King, I could not have asked for a better mentor and am

forever grateful for your faith in allowing me to develop and lead the beehive fence trial in Sri Lanka.

There were many hurdles along the way: some I was prepared for but many for which I was not. This

project opened my eyes to many facets of life in conservation and your support was invaluable both

professionally and personally. I extend my gratitude to the entire Save the Elephants and Elephants

and Bees teams – it’s been an honour.

To my absolute superstar of a research assistant Supun Herath: you joined our team on day one and I

don’t know what I would have done without you. Thank you isn’t enough to express my appreciation

for your endless hard work, your English-Sinhalese translations, for putting up with the beestings and

constantly fixing fences, for keeping me safe from the elephants, and for letting Arrow in your tuk tuk.

A huge thank you also to Sarath and Siriya for always being available to help with anything. The three

of you were my A-Team and now lifelong friends: I miss our sunset hangs on the rock.

Many thanks to all from the Sri Lankan Wildlife Conservation Society who contributed, especially

Chathuranga Dharmarathne for your support during a particularly challenging field season, and to

Swarna and Leela for your endless smiles, curries and kindness towards me and my posse of village

dogs. Your friendship is always in my heart. I am also extremely grateful to local and international

beekeepers Dr. R.W.K. Punchihewa, Mr. Frank Ryde, Dr. David Gormley-O’Brien and Ms. Emma Collins.

Supun and I learned so much from your expertise, as did the Dewagiriya community. David and Emma,

that you came all the way from Australia to help a complete beekeeping novice, on a project you’d

just heard of, is beyond amazing.

Thank you also to my fantastic team of interns who helped with tasks big and small at various stages

of the project: Lauren Ross, Lorna McCallister, Manon van der Meer and Louise Jory. You guys slogged

it out through torrential rains, intense heat, charging elephants, sleeping in leaky rooms, power

outages, snakes and scorpions. And to those who took it one step further and came out to lead the

field team after I returned to Australia: Jessica van Fleteren, Rebecca Sargeant, Robin Cook and Zaineb

Akbarally. You often had to put up with far more than you’d bargained for and I am so grateful for all

your hard work. Special extra thanks to intern and project leader Brianna Stoyle who had probably the

hardest job of all wrapping up our final field season. All of you are absolute gems.

I am also thankful for the support, practical assistance, and company of new and old friends from

around the globe. Jo Heaton, Simon King, Jinalee de Silva, Hollis Burbank-Hammarlund, Yufei Gao,

Sarah Reid-Anderson and Natalie Gonzalez – thank you for helping in the field, for the great

conversations, and the sunset Lions. To friends who visited me on-site and often found themselves

digging holes and taking notes – I appreciate it immensely. That means you: Adam ‘Squire’ Mattinson,

James Ahern, Tim ‘BC’ Hunt, Lynsey Humes, Alistair ‘Stunt’ Hunt, and Joel Graham.

To the beautiful Dewagiriya community, I know you won’t read this but I want send my heartfelt

gratitude out to the universe. I was a stranger introducing a seemingly crazy idea about bees and

elephants, and I was welcomed warmly. Thank you to everyone who shared their stories and

experiences, who built and monitored fences, and for the endless cups of sugary black tea. I have no

words for how lucky I feel to have met you all. Special thanks to the Sobana, Jaysinghe, Sirisena,

Somathilaka, Ariyadasa, Guneris, Somanawathi, Dharmadasa, Ghanawathi, Karunadasa, and Jayasena

families for participating in the experimental trial. I wish you all happiness in your families, success in

your harvests, and safety living in proximity to the elephants.

A big shout-out to my brainy ‘cuz’ Nicky Butler who let me rant and rave, read her excellent thesis for

guidance, and found me an awesome statistics tutor. And to Alex Masey – without you, my data would

still be sitting in excel spreadsheets. Thank you for making ‘R’ less scary and for your patience and the

time you took away from your own work to help me out.

There are only a few people who truly ‘get’ why someone wants to gallivant around the globe chasing

elephants and bees and living a never-ending student life. Everyone mentioned here understood

something unique to this experience: either what it’s like to work in this industry or why I sometimes

have to do things the way I do: Tamsin Webb and Sundari Sitaram, there are some things that can only

be truly understood by people doing something along the same lines and I am so glad we connected.

Kirby Smith, Camille Royal and Susan Gorecki thank you for listening to my elephant tales and helping

with some tough decisions. Sarah Leatham and Dianne Butler, thanks for visiting me in Sri Lanka, for

the wine dates, and for ‘getting’ me.

To my husband Morgan Tipper, thank you for everything: for coming out to Sri Lanka and doing

whatever was needed in the field, for your technical mapping and computer assistance, for supporting

us financially, for your patience with having an often-absent partner and wife, for never getting tired

of me blabbing on about my work, and most importantly for always being on my side. Whether

intended or not, this adventure has been as much yours as mine. Let’s see what’s next!

Finally, I would like to dedicate this project to the late Mr. Jayasena and his wife Mrs. Jaysena. Both

respected elders in Dewagiriya Village, Mr. Jayasena passed away before his time in 2016 from Chronic

Kidney Disease (CKD). Sadly, many community members are afflicted by this due to contaminated

water sources. Their joint support of Mr. and Mrs. Jayasena and their enthusiasm for the beehive

fence project was instrumental in making it a reality. Their smiles and shared stories will remain in my

heart forever.

Mrs. Jayasena holding a photo of her

husband and I next to their beehive

fence. © Zaineb Akbarally

Abstract

Human-elephant conflict (HEC) in the form of crop-raiding, is a major conservation challenge to the

long-term survival of elephant populations, simultaneously threatening the livelihoods and personal

safety of people living in proximity to elephants. The widespread problem of HEC has led to a great

deal of research into the causes, consequences and predictors of elephant crop-raiding activity.

However, despite similarities across HEC situations, site-specific differences are also apparent.

Furthermore, most studies focus on one facet of HEC when it is a complex issue requiring

understanding of local elephant behaviour, identification of the characteristics and patterns of crop-

raiding at the local scale, and careful implementation and monitoring of mitigation strategies.

In this study, I selected a region of Sri Lanka experiencing high levels of HEC and sought to provide an

in-depth assessment of the site-specific situation generated over a three-year period. Specifically, we

aimed to: identify general patterns of behaviour occurring in local areas representing differing levels

of anthropogenic disturbance to elephants; profile patterns and predictors of crop-raiding activity in

a village heavily impacted by HEC; and test the effectiveness of beehive fencing as an Asian elephant

deterrent tool.

First, I provide initial evidence that elephants inhabiting areas of ‘medium’ level anthropogenic

disturbance outside of protected boundaries, interrupt feeding and increase ‘reactive’ behaviours

such as smelling and holding the ‘vigilance’ posture in response to immediate anthropogenic threats

in the environment. In the absence of any known disturbances, there was no difference in general

behaviours of male or female elephants between the two risk zones. Secondly, I show that elephant

crop-raiding in Dewagiriya Village occurs year-round and follows no clear seasonal patterns. Similar to

other HEC situations, male elephants are the predominant crop-raiders, and crop-raiding occurs

almost exclusively at nights. Within-site variations in crop-raiding intensity were also identified, with

properties closest to water tanks and forest habitat the most vulnerable. Finally, our three-year

beehive fence trial showed that households using beehive fences around their gardens had

significantly less elephant visits into their gardens then households without. Still, difficulties in

attracting natural colonies, poor honey production, set-up costs, and farmer motivations were barriers

to success.

This study contributes to the general body of knowledge on elephant behaviour in anthropogenically

influenced contexts, and specifically on patterns of crop-raiding and mitigation efforts. Further

research into the potential of beehive fences as an Asian elephant deterrent, preferably in a location

more amenable to beekeeping, would help to determine the value of expanding this technique further

in Sri Lanka, and elsewhere in Asia.

1

Chapter 1: Introduction

2

1.1 Human-wildlife conflict

Human-wildlife conflict (HWC), where the actions of wildlife negatively impact the well-being of

humans or vice versa, is an ongoing and increasing global environmental concern (Manfredo 2015;

Baruch-Mordo et al. 2008; Madden 2004). An array of wildlife from rodents to large mammals, are

implicated in HWC (Harich et al. 2013; Nijman and Nekaris 2010; Woodroffe et al. 2005b; Gillingham

and Lee 2003), encompassing varying situations from squirrels eating crops (Harich et al. 2013), to

wild dogs predating on livestock (Woodroffe et al. 2005a), to bears posing threats to park-goers (Lewis

et al. 2015; Baruch-Mordo et al. 2008). On the other side of the coin, humans often drive conflict by

hunting or poaching, such is the case with the illegal killing or capturing of wild animals for 'bush meat'

(Ripple et al. 2016) or private pet trades (Menard et al. 2013).

Wildlife entering farmlands to forage on crops, commonly known as 'crop-raiding', has been a

prominent HWC concern for decades and remains topical with no signs of abating (examples: Saraswat

et al. 2015; Hoffmeier-Karimi and Schulte 2014; Thapa 2010; Gillingham and Lee 2003; Bandara and

Tisdell 2002; Rao et al. 2002; Studsrod and Wegge 1995; Sukumar 1990). Communities of lower socio-

economic status, particularly those who live on the fringe of traditional wildlife habitat and rely on

subsistence farming, are often disproportionately affected by crop-raiding (Barua et al. 2013; Ogra

2008; Rao et al. 2002), and many animals have been described, fairly or otherwise, as 'agricultural

pests' including rodents (e.g. squirrels, Sciuridae spp: Harich et al. 2013), primates (e.g. rhesus

macaques, Macaca mulatta, Saraswat et al. 2015; and baboons, Papio anubis and Papio cynocephalus:

Hill 2000), ungulates (e.g. barking deer, Muntiacus muntjak: Rao et al. 2002, and peccaries, Tayassu

tajacu: Perez and Pacheco 2006) and large terrestrial mammals (e.g. elephants, Loxodonta africana,

Hoffmeier-Karimi and Schulte 2014; Elephas maximus: Santiapillai et al. 2006; and rhinoceros,

Rhinoceros unicornis: Thapa 2010).

Injury or death to both humans and wildlife is a serious consequence of HWC involving crop-raiding

(Barua et al. 2013; Choudhury 2004; Rao et al. 2002). Additionally, substantial economic losses (e.g.

when property or crops are damaged) and social costs (e.g. interruption of schooling or work due to

the presence of potentially dangerous wildlife) can have ongoing deleterious effects on people (Barua

et al. 2013; Harich et al. 2013; Hartter et al. 2011; Ogra 2008; Thirgood et al. 2005; Hill 2000). As

humans struggle to cope with consequences of HWC, negative perceptions of co-existence with

3

certain wildlife species grow (Dickman 2010; Madden 2004). Although not always responsible for the

most serious damage, species of a larger size which typically present a higher level of perceived

physical danger to humans, tend to receive the least tolerance and elicit harmful retaliatory actions

from people (Hoffmeier-Karimi and Schulte 2014; Ogra 2008; Naughton-Treves and Treves 2005; Hill

2004; Bandara and Tisdell 2002).

With the continuing growth of human populations, and expansion of communities into land inhabited

by wildlife, it seems inevitable that human-wildlife conflicts will only increase (Manfredo 2015; Harich

et al. 2013; Redpath et al. 2013; Fernando et al. 2005; Madden 2004). As a result, the conservation

efforts and long-term survival of several endangered and iconic wildlife species are being threatened

(Graham et al. 2010; Naughton-Treves and Treves 2005; Loe and Roskaft 2004; Thirgood et al. 2005;

Madden 2004). A prime example of this is the conflict involving humans and both African and Asian

elephants (Graham et al. 2010; Perera 2009; Sitati and Walpole 2006; Bandara and Tisdell 2002).

1.2 Human-elephant conflict

There are various forms of conflict involving humans and elephants, the two most common being the

illegal poaching of elephants to harvest tusks for the ivory trade, and crop-raiding by elephants.

Poaching primarily occurs in Africa where both male and female elephants typically have tusks (Ihwagi

et al. 2018; Burn et al. 2011; Kahindi et al. 2009), whereas crop-raiding is a major HEC issue in almost

all African and Asian elephant range countries (Okello 2016; Davies et al. 2011; Webber et al. 2011;

Graham et al. 2010; Perera 2009; Sitati and Walpole 2006; Sitati et al. 2003). Both are major

conservation challenges to the long-term survival of elephant populations however it is the issue of

crop-raiding that forms the focus of this study.

Although African and Asian elephants are distinct species, with many unique morphological and

behavioural characteristics, there are several fundamental characteristics of elephant crop-raiding

that transcend locations, countries and continents. In most locations, the primary drivers of HEC are

the loss or fragmentation of traditional elephant habitat and disruption of elephant movement

patterns, mainly due to deforestation, agricultural expansion and human encroachment into, or very

4

close to, elephant inhabited areas (Amwata and Mganga 2014; Chakraborty and Mondal 2013;

Campos-Arceiz et al. 2009; Fernando et al. 2005; Choudhury 2004; Sitati et al. 2003). Common

consequences of HEC are personally significant economic losses to farmers when crops are eaten or

trampled (Okello et al. 2016; Amwata and Mganga 2014; Gubbi 2012; Ekanayaka et al. 2011; Fernando

et al. 2005), damage to property as elephants attempt to access crops stored inside (Campos-Arceiz

et al. 2009), injury or death to humans and elephants (Chakraborty and Mondal 2013; Gubbi 2012;

Choudhury 2004; Sitati et al. 2003), disturbed sleep and work patterns as people are required to

continuously guard crops (Barua et al. 2013; Hill 2004), and negative impacts on human physical and

mental health (Barua et al. 2013; Ogra 2008; Sukumar 2006; Hoare 2000). Unsurprisingly, HEC fuels

negative perceptions of co-existing with elephants and hinders local conservation efforts (Amwata

and Mganga 2014; Chakraborty and Mondal 2013; Fernando et al. 2005).

Several common factors that influence the likelihood of elephants raiding crops have also been

identified across multiple sites. These include the proximity of villages and farms to elephant habitat

(Barnes et al. 2015; Gubbi 2012; Graham et al. 2010; Sitati et al. 2003), the types of crops grown and

the growth stage of crops, with ripening maize and rice particularly susceptible (Chakraborty and

Mondal 2013; Gubbi 2012; Webber et al. 2011; Graham et al. 2010; Campos-Arceiz et al. 2009; Chiyo

et al. 2005), and seasonal changes and rainfall (Gubbi 2012; Webber et al. 2011; Campos-Arceiz et al.

2009; Fernando et al. 2005). It is typical for male elephants to raid crops more frequently than female

elephants (Thaufeek et al. 2014; Das et al. 2014; Graham et al. 2010) and crop-raiding occurs almost

exclusively at nights (Das et al. 2014; von Gerhardt et al. 2014; Graham et al. 2010).

Despite the aforementioned commonalities across sites, it is wise to exercise caution when making

generalisations, even if the crop-raiding sites in question are in the same country or the same area of

a given country, and particularly when making management or mitigation decisions.

1.2.1 Human-elephant conflict in Sri Lanka

Sri Lanka is a country with a long and complicated history of human-elephant interactions. Although

elephants are traditionally revered as a cultural and religious icon (Fernando et al. 2011; Santiapillai

5

et al. 2006; Bandara and Tisdell 2003), wild elephants have historically been subjected to high levels

of negative interactions with humans. From the hunting and capture of elephants in the early 1900's

(Agar n.d; Hennessey 1929), to being caught in the crossfire of a lengthy and violent civil war

(Santiapillai et al. 2006; de Silva 1998; Alahakoon and Santiapillai 1997), to deforestation and habitat

loss (Haturusinghe and Weerakoon 2012; Wikramanayake et al. 2004), the relationship between

humans and elephants in Sri Lanka has been fraught with conflict. Crop-raiding has been discussed in

popular media since the mid 1900's (Hennessey n.d; Tutein-Nolthenius n.d) however it was around

the mid 1990's when it began to gain global attention as a serious elephant conservation issue (de

Silva 1998; Santiapillai 1996; Santiapillai and de Silva 1994). In 2005, HEC was recognised as the

principal threat to the survival of Sri Lanka's wild elephant population (Fernando et al. 2005) and it

continues to be a major issue in almost all elephant inhabited areas of Sri Lanka (Fernando et al. 2019;

Santiapillai et al. 2010) (Figure. 1.1).

Figure 1.1: Sri Lanka: National Parks and elephant distribution © Morgan Tipper

6

As in other locations where crop-raiding occurs, the disruption of traditional elephant habitat and

ongoing expansion of agriculture closer to forests perpetuate interactions between humans and

elephants (Haturusinghe and Weerakoon 2012; Fernando et al. 2011; Campos-Arceiz et al. 2009;

Santiapillai 1996). Specifically, large-scale development including major damming projects for

irrigation and the resettlement of people near forest boundaries, are bringing humans and elephants

into direct competition for resources (Fernando et al. 2011; Santiapillai et al. 2006; de Silva 1998).

It is estimated that more than 200 elephant (Fernando et al. 2011) and 50 human (Campos-Arceiz et

al. 2009; Perera 2009) deaths occur annually in Sri Lanka as a consequence of HEC. Non-fatal injuries

to elephants and humans (Thaufeek et al. 2014; Campos-Arceiz et al. 2009; Wikramanayake et al.

2004), crop loss (Thaufeek et al. 2014; Haturusinghe and Weerakoon 2012; Campos-Arceiz et al. 2009;

Wikramanayake et al. 2004), property damage (Thaufeek et al. 2014; Haturusinghe and Weerakoon

2012; Campos-Arceiz et al. 2009), and disruption of people's regular activities due to crop-guarding

efforts and being unable to safely move about the village when elephants are present, are further

serious ramifications (Santiapillai et al. 2010; de Silva 1998).

Elephants raid a wide variety of fruits, vegetables and grains in Sri Lanka during all seasons

(Haturusinghe and Weerakoon 2012; Ekanayaka et al. 2011; Campos-Arceiz et al. 2009;

Wikramanayake et al. 2004) however major crops such as rice are prime attractants and crop-raiding

often peaks as rice plants mature (Thaufeek et al. 2014; Haturusinghe and Weerakoon 2012; Fernando

et al. 2011). As with other areas experiencing high HEC throughout Asia, elephant crop-raiding

behaviour is almost entirely nocturnal (Campos-Arceiz et al. 2009; Wikramanayake et al. 2004;

Bandara and Tisdell 2003) and it is more common for male than female elephants to enter farmlands

to crop raid (Thaufeek et al. 2014; Haturusinghe and Weerakoon 2012; Ekanayaka et al. 2011; Bandara

and Tisdell 2002).

Most of the villages in close proximity to elephants are home to low-income farming families reliant

on annual or bi-annual harvest of rice crops for subsistence and income generation (Santiapillai et al.

2010; de Silva 1998). Ongoing crop-raiding activity increases desperation of communities to protect

their livelihoods, and people resort to more and more violent methods to deter elephants

(Haturusinghe and Weerakoon 2012; Santiapillai et al. 2010). Consequently, elephants respond more

aggressively to human presence, and a cycle of conflict ensues (Fernando et al. 2011). Despite

substantial efforts by conservationists and researchers to understand and mitigate crop-raiding in Sri

Lanka, it remains a prominent issue.

7

1.3 Characteristics and behaviour of elephants in high human-elephant conflict areas

The widespread problem of HEC has led to a great deal of research into the causes, consequences and

influencing predictors of elephant crop-raiding activity. However, for many locations experiencing

high levels of HEC, less is known about the general behaviour patterns (e.g. movement and foraging

habits, and responses to disturbance) of the overall local elephant population, especially in proximity

to farms and villages.

Knowledge of even relatively simple demographic characteristics can be hazy. For example, the best

available estimate of elephant abundance in Sri Lanka comes from a 2011 census conducted by the Sri

Lankan Department of Wildlife and Conservation (DWC). This census produced a population count of

5,879 wild elephants (Fernando et al. 2011; The Guardian 2011). However, it is likely population figures

have changed since then.

This lack of knowledge of precise elephant abundance near farms and villages clearly impedes wildlife

management planning. Only recently has a country-wide survey been completed revealing that

elephant range extends over almost 60% of Sri Lanka with people residing in almost 70% of this range

and conflict (especially involving male elephants) occurring almost everywhere humans and elephants

co-exist, (Fernando et al. 2019). This aligns with prior research identifying almost 65% of suitable

elephant habitat as existing outside of designated protected areas (PA’s) (Santiapillai et al. 2006) and

positing that the majority of elephants in Sri Lanka spent some or all of their time outside of PA’s

(Weerakoon et al. 2004).

In the absence of precise population figures and thus accurate estimates as to how many elephants

are involved in conflict, generating demographic and behavioural information of elephants utilising

habitat near farms and villages can provide valuable insight for predicting future crop-raiding activity

or even how elephants might react to deterrents or human presence.

1.3.1 Risk taking behaviour

Essentially, elephant crop-raiding is a high-risk, high-gain foraging strategy whereby elephants leave

the relative safety of PA's or forest habitat to enter farmlands and feed on crops. This foraging strategy

8

puts elephants in close proximity to humans. Given that people scare, injure or even kill elephants to

deter them from farms, interactions with humans present a substantial threat (Evans and Adams

2018). On the other hand, elephants may see proximity to human inhabited areas as a source of

opportunity, where nutritious fields of poorly guarded or unguarded crops are readily available (Evans

and Adams 2018).

Several behaviours of wild animals, including foraging strategies, are altered due to increasing human-

animal interactions, particularly conflict events (Fischhoff et al. 2008; Lacy and Martins 2003; Anthony

and Blumstein 2000; Olson et al. 1997). In three separate examples, lizards habituating to areas of

high human use showed more aggression than lizards in low human use areas (Lacy and Martins 2003),

Alaskan brown bears altered seasonal foraging patterns according to their degree of habituation to

humans (Olson et al. 1997), and zebra reduced use of grassland habitat when lions were observed in

the area on the same day (Fischhoff et al. 2008).

Crop-raiding, where animals make the decision to enter a potentially risky area to forage, is often

characterised by strong male-biased sex differences in frequency and intensity of raiding activity, with

males typically raiding more frequently, for longer durations, and causing more damage than females

(e.g. chimpanzees: Wilson et al. 2007; vervet monkeys: Saj et al. 1999). This is certainly true of

elephants. While crop-raiding is not exclusively a male activity, males display a much greater

propensity to exploit this foraging strategy then do females, even though all elephants within the

population could theoretically access crops (de Silva et al. 2013; Haturusinghe and Weerakoon 2012;

Fernando et al. 2011). Furthermore, in areas where both sexes do crop-raid, male elephants tend to

raid year-round while females limit crop-raiding activity to times when crops are nearing maturation

and to locations closest to forest boundaries (Williams et al. 2001).

In one of the few in-depth studies investigating the impacts of risk on Asian elephant behaviour,

Srinivassaiah et al. (2012) used areas of varying degrees of human disturbance to represent low to

high levels of risk, and monitored elephant behaviour patterns in each area. It was found that time

spent in-and outside of PA's, time spent feeding and specific behaviours all reflected the level of

perceived risk present. For example, elephants spent most of their time foraging or moving. However,

as the level of human disturbance (risk) became higher, time spent foraging decreased but time spent

moving increased as did vigilance behaviours. Adult male elephants spent more time in high risk areas

than any other elephants, demonstrating sex and age-based influences on risk-taking behaviour for

this population (Srinivassaiah et al. 2012).

9

Other studies relating to behaviour in the presence of risk have shown that elephants possess the

capacity to assess varying levels of risk and adjust their behaviour accordingly. African elephants used

olfactory and visual cues to distinguish between the perceived presence of men from the Masai tribe

(who represent a threat to elephants) and men from the Kamba tribe (who do not represent a threat

to elephants) and fled from the Masai stimuli (Bates et al. 2007). Similarly, Asian elephants exposed

to auditory cues of tiger growls (representing a real threat) with immediate silent retreat, and to

leopard growls (representing no real threat) with more investigative behaviours rather than avoidance

(Thuppil and Coss 2013).

With reference to elephant foraging behaviour in areas of high human presence, it could be that

females adjust their behaviour to the perceived level of risk by spending more time in PA's and

avoiding high-risk areas, whilst males fine-tune specific behaviours, moving faster across human-

dominated landscapes, entering farms under the cover of darkness, and selecting crop fields with the

least guarding effort present (Evans and Adams 2018; Graham et al. 2009).

Identifying behavioural differences of elephants in areas of varying degrees of human-elephant

interaction and potential influences of sex, age, and prior experience with humans can help to predict

the behaviour of individuals, or groups of individuals, across time and contexts (Freeman et al. 2013).

1.4 Human-elephant conflict mitigation

For as long as elephants have been raiding crops, farmers have been utilising a wide array of mitigation

methods with varying degrees of success. Since the 1990's, much research has focused on the

development, monitoring and assessment of elephant deterrent techniques (e.g. Africa: Hoare 2015;

Graham 2010; Graham and Ochieng 2008; Sitati and Walpole 2006; Dublin and Hoare 2004; Osborn

and Parker 2003; Asia: Gunaryadi et al. 2017; Thirunavukarasu 2014; Davies et al. 2011; Thapa 2010;

Perera 2009; Fernando et al. 2008; Wikramanayake et al. 2003; de Silva 1998; Santiapillai 1996).

Mitigation efforts can broadly be categorised as large-scale (e.g. methods implemented at

government level), traditional (e.g. methods that have long been used by individual farmers), or

community-based (e.g. methods implemented at a village level, often with the help of NGO's or other

research/conservation organisations). Most deterrents have been tried at numerous locations,

spanning both Africa and Asia (Hoare 2015; Fernando et al. 2008; Nelson et al. 2003). However, despite

10

a long history of tried and tested HEC mitigation methods, crop-raiding levels show no sign of abating.

What is becoming increasingly apparent is that there is no 'one size fits all' solution and that a

combination of deterrents based on site-specific knowledge of crop-raiding activity, the local elephant

population, and community perceptions may be the best approach (van de Water and Matteson 2018;

Pozo et al. 2017; Karidozo and Osborn 2015; Fernando et al. 2008).

Crop-raiding mitigation methods commonly implemented at a government level include confinement

of elephants into PA's through elephant drives, translocation of problem individuals, and large-scale

electric fencing (van Eden et al. 2016; Ponnusamy et al. 2016; Fernando et al. 2011; Santiapillai et al.

2006; Nelson et al. 2003). These methods are expensive and difficult to maintain and often do not

take into account the behaviour and movement patterns of elephants (Fernando et al. 2011; Perera

2009; Santiapillai 1996). Farmers are often encouraged to create 'elephant barriers' such as digging a

trench (Fernando et al. 2008) or planting unpalatable crops (Gross et al. 2017; 2016; Fernando et al.

2008; Nelson et al. 2003). Trenches have a high failure rate as they are generally either prohibitively

expensive or not strong enough to withstand elephant encroachment (Fernando et al. 2008). Planting

unpalatable crops has its merits and some success has been shown (Gross et al. 2017; 2016; Fernando

et al. 2008), however convincing farmers to change their cultivation habits is difficult and may not

always be economically viable (Fernando et al. 2008).

'Traditional' small-scale methods commonly employed by farmers include hanging tin cans from wire,

guarding crops from a tree hut or watch tower at night, making loud noises, shining bright flashlights,

and throwing firecrackers or fire sticks (sticks with an oil soaked rag set alight at one end) in the

direction of elephants (van de Water and Matteson 2018; Haturusinghe and Weerakoon 2012;

Fernando et al. 2011; Hedges and Gunayardi 2009; Perera 2009). Over time elephants habituate to, or

find means of overcoming, most of these methods (Fernando et al. 2011; Osborn and Parker 2003).

Many farmers struggle to protect their crops using only traditional methods, and over time may resort

to using increasingly harmful means to deter elephants. Some farmers set trap guns, plant live electric

wires, hide poison in crops, or use shot guns or explosives to injure or kill elephants (Haturusinghe and

Weerakoon 2012). As humans increase their aggressiveness towards elephants, elephants appear to

be adjusting their behaviour by reacting more aggressively towards humans. Thus, a spiral of HEC

escalation ensues (Fernando et al. 2011; Perera 2009).

Over the last two decades, a shift towards community-based mitigation encouraging farmers to use

non-violent methods has gained traction (van de Water and Matteson 2018; Changa et al. 2016;

11

Karidozo and Osborn 2015; Hedges and Gunaryadi 2009; Graham and Ochieng 2008; Sitati and

Walpole 2006). Ideally, community-based deterrents are relatively cheap, require little time to

establish and maintain, and encourage ownership and responsibility by the farmer (Changa et al. 2016;

Karidozo and Osborn 2015; King et al. 2011; 2009). Effectiveness of a community-based deterrent

depends not only the techniques ability to deter elephants but on how feasible it is for farmers to

implement and maintain, and their willingness to adopt the technique (King et al. 2011; 2009). Some

success has been demonstrated with initiatives using chilli based deterrents (Botswana: Pozo et al.

2017; India: Baishya et al. 2012; Sumatra: Hedges and Gunaryadi 2009; Tanzania: Changa et al. 2016),

bio-barriers of unpalatable crops (Nepal: Gross et al. 2017; Zambia: Gross et al. 2016), and beehive

fencing (Kenya: King et al. 2017; 2011; India: Nair and Jayson 2016) however further research and trials

are required to properly understand where and when these methods may be appropriate.

1.4.1 Beehive fencing

Beehive fencing is a relatively new community-based mitigation method which has shown promise in

Africa (King et al. 2017, 2011; Scheijen et al. 2018), and is gaining popularity not only for its properties

as an elephant deterrent but also due to recent worldwide concerns about the collapse of honey bee

colonies and the global push for promoting beekeeping and honey bee health (Meixner and Conte

2016). This novel solution came about in the early 2000's when researchers from conservation

organisation Save the Elephants (Kenya) discovered that African elephants avoided resting under trees

containing African honey bees (Apis mellifera scutellata) (Vollrath and Douglas-Hamilton 2002). This

was followed by a series of experiments where elephants were exposed to audio playback of bee

sounds and responded by rapidly retreating from the sounds, performing exploratory (e.g. smelling)

or risk avoidance (e.g. shaking heads to potentially keep bees away from sensitive areas) behaviours,

and producing rumble vocalisations (King et al. 2007). These rumble vocalisations were recorded and

played back to elephants who again responded with retreat and/or exploratory and avoidance

behaviours. This indicated that not only were elephants aware of the danger presented by the

perceived presence of bees, but that elephants could communicate this threat to conspecifics who

also responded accordingly (King et al. 2010).

Based on the concept that crop-raiding could be reduced by capitalising on the bee-avoidance

behaviour elephants exhibited during playback experiments, the 'beehive fence' was designed (King

12

et al. 2011; 2009). A beehive fence is a series of beehives hung from posts and connected to one

another by wire linking one beehive to the next, with beehives approximately 8 m apart (Figure. 1.2).

Elephants attempting to enter the area inside the beehive fence, will hit the connecting wire between

beehives, causing the attached beehives on either side to swing and, if occupied, disturb the bees

causing them to exit the hive and swarm at the elephants (King et al. 2011).

Figure 1.2: Example of a beehive fence: from Dr. Lucy King’s ‘Elephants and Bees’ research site in Sagalla, Kenya

© Kylie M. Butler

Today in Africa, beehive fencing is being utilised as an effective, relatively inexpensive, and low-

maintenance elephant deterrent, capable of deterring elephants from entering the fence perimeter

for up to 80% of approaches (King et al. 2017). In addition to providing protection of crops against

elephant depredation, farmers derive further benefits as honey and other bee products can provide

an additional income source or be kept for personal use, and honey bees provide pollination services

(King et al. 2017; 2011; 2009). Following on from initial success at multiple locations in Kenya (King et

al. 2011; 2009), results from a beehive fence trial in Tanzania using a linear fence to divide the park

boundary from nearby farmlands, also showed the potential of beehive fencing as a crop-raiding

13

mitigation tool (Scheijen et al. 2017). Additionally, beehives are being used in South Africa (Cook et al.

2017) and Gabon (Ngama et al. 2016) to protect valuable trees from elephant damage.

The success of beehive fencing in Africa, has led to questions about its potential applicability as an

elephant deterrent in Asia. Although there are many locations across the two continents with similar

crop-raiding characteristics, the species of elephant differs and more importantly, notable behavioural

differences exist between the African honey bee and Asian honeybee (Apis cerana cerana). Of most

concern is the difference in aggression, with African honey bees known to sting much more readily

and to swarm for longer distances when disturbed (Winston 1987), while Asian honey bees are

comparatively more placid (Punchihewa 1994). However, initial audio playback experiments testing

the response of wild elephants in Sri Lanka to the perceived presence of Asian honeybees, suggest

that avoidance responses are typical, although not as marked as in their African counterparts (King et

al. 2018).

To date, the only existing evidence of beehive fencing exhibiting success as an Asian elephant

deterrent is presented in two anecdotal newspaper reports from India (Dutta 2013; Manoj 2012), and

one three-month research trial, also in India where beehives were arranged in a short linear formation

blocking elephant pathways to fields from a nearby forest (Nair and Jayson 2016). Although this trial

showed promise, only 14 elephant approaches to the fence with two breaches were recorded and an

individual male elephant was responsible for ten of these approaches. Longer term studies over

multiple seasons are required to measure how the beehive fence performs over time and how

individual elephants respond to it, before it can be accurately assessed as a successful deterrent or

otherwise.

1.5 Rationale

Although extensive research exploring various facets of HEC from a multitude of sites exists, the vast

majority of studies focus on particulars: for example, crop-raiding patterns, or demographic

characteristics of the local population, or a specific mitigation technique. Even when multiple aspects

of the conflict situation are examined, knowledge of the behaviour patterns of the local elephant

population is often lacking. Information from these studies makes a valid contribution to the HEC body

of knowledge, however the importance of generating as complete a picture as possible of site-specific

14

conflict situations cannot be understated. This is particularly so when deciding on mitigation

techniques to implement or when recommending specific mitigation techniques based on success or

failure at a given location.

1.6 Thesis overview

In this study, I selected a region of Sri Lanka experiencing high levels of HEC in the form of crop-raiding

(see Chapters 2 and 3 for detailed site descriptions) and sought to provide an in-depth assessment of

the site-specific situation generated over a three-year period. The overarching objectives of this study

were:

a) To describe the general behaviour patterns of an elephant population in a high HEC

region of Sri Lanka, both inside and outside of the local protected area, Wasgamuwa

National Park (WNP)

b) To create a profile of the crop-raiding situation in a village near to WNP, including

identifying predictors of crop-raiding activity

c) To test the effectiveness of a new and potential HEC mitigation technique for Asia –

beehive fencing, and evaluate its overall potential as an Asian elephant deterrent

In Chapter 2, I use video footage of members of the local elephant population captured over two years

of observations to score general behaviour (e.g. feeding and small-scale movement patterns) and

specific actions (e.g. smelling and head-shaking) from a pre-determined ethogram. I compare these

behaviours and actions for elephants inside and outside of WNP, and also in the presence and absence

of known disturbances (e.g. approach of motorbike, tuk tuk or other vehicle). The potential influence

of sex and prior level of negative interaction with humans (using wounds/scars as a proxy) on

behaviour was also examined.

In Chapter 3, I move into the centre of the conflict zone: a small village on the outskirts of WNP called

Dewagiriya, which typically experiences high levels of crop-raiding year-round. Before implementing

mitigation techniques, it is vital to have a sense of the patterns and influencing factors of crop-raiding

15

activity, and this has been described for many crop-raiding sites in both Africa and Asia. Here, I use

farmer collected data and farmer interviews to determine seasonal, night/day, and lunar patterns of

crop-raiding, as well as typical group size and sex of crop-raiders (if known) and damage incurred. I

look at the potential influences of crop-type and farm proximity to forest habitat on susceptibility to

elephant encounters. These factors are commonly examined in crop-raiding studies and incorporating

them into this study facilitates meaningful comparison across sites. Additionally, I placed night-vision

camera traps in strategic locations throughout the village and in forest habitat on the periphery to

capture footage of elephants utilising these areas with the aim of identifying specific individuals, or at

least their gender and age class.

Finally, in Chapter 4, I present findings from a three-year beehive fence trial in Dewagiriya village,

assessing the effectiveness of beehive fences by comparing elephant events and entry into home

garden areas for experimental (beehive fence) and control (non-beehive fence) properties during the

same periods. Assessing deterrent effect is an important element of determining the success of any

mitigation technique however there are multiple factors essential to evaluating overall feasibility of a

deterrent tool, many of which are commonly overlooked in the literature. I incorporate not only the

deterrent effect of beehive fences but community perceptions of its usefulness and their willingness

to participate; cost and time analysis of set-up and maintenance; and benefits and challenges of

beekeeping, in an overall assessment of its potential as an Asian elephant deterrent tool in Sri Lanka.

Chapter 5 concludes this study by providing a discussion on our findings from the aforementioned

chapters, including the benefits of a multi-faceted approach that simultaneously investigates the

behaviour of the local elephant population, with in-depth analysis of crop-raiding patterns and

influencing factors, and a comprehensive assessment of a newly implemented mitigation technique.

We discuss the limitations and challenges of the study, provide recommendations as to the future

potential of beehive fencing as an Asian elephant deterrent, and suggest areas where further research

could be conducted.

16

1.7 References

Agar, RS n.d. 'Proclaimed elephants', IN: (eds: Ranasinghe, D, Ratnapala, S, Seneviratne, L,

Schwammer, H, Schwammer, G and Perera, BH) Saga of the elephant in Sri Lanka: Extracts from Loris

Volume 1 (1936) to Volume 10 (1964), Nedimala, YN Print Graphics, pp. 17 - 19.

Alahakoon, J & Santiapillai, C 1997, 'Elephants: unwitting victims in Sri Lanka's civil war', Gajah, vol.

18, pp. 63 – 65.

Amwata, DA & Mganga, KZ 2014, 'The African elephant and food security in Africa: experiences from

Baringo District, Kenya', Pachyderm, vol. 55, pp. 22 – 28.

Anthony, LL & Blumstein, DT 2000, 'Integrating behaviour into wildlife conservation: the multiple ways

that behaviour can reduce Ne', Biological Conservation, vol. 95, pp. 303 – 315.

Baishya, HK, Dey, S, Sarmah, A, Sharma, A, Gogoi, S, Aziz, T, Ghose, D & Williams, AC 2012, 'Use of chilli

fences to deter Asian elephants – a pilot study', Gajah, vol. 36, pp. 11 – 13.

Bandara, R & Tisdell, C 2003, 'Comparison of rural and urban attitudes to the conservation of Asian

elephants in Sri Lanka: empirical evidence', Biological Conservation, vol. 110, pp. 327 – 342.

Bandara, R & Tisdell, C 2002, 'Asian elephants as agricultural pests: economics of control and

compensation in Sri Lanka', Natural Resources Journal, vol. 42, pp. 491 – 519.

Barnes, RFW, Danquah, E, Hema, EM, Dubiure, UF, Manford, M, Nandjui, A & Boafo, Y 2015,

'Retrospective versus prospective designs for studies of crop-raiding by elephants in Kakum, Ghana',

Pachyderm, vol. 56, pp. 44 – 50.

Barua, M, Bhagwat, SA & Jadhav, S 2013, 'The hidden dimensions of human-wildlife conflict: health

impacts, opportunity and transaction costs', Biological Conservation, vol. 157, pp. 309 – 316.

Baruch-Mordo, S, Breck, SW, Wilson, KR & Theobald, DM 2008, 'Spatiotemporal distribution of black

bear-human conflicts in Colorado, USA', Journal of Wildlife Management, vol. 72, no. 8, pp. 1853 –

1862.

17

Bates, LA, Sayialel, KN, Njiraini, NW, Moss, CJ, Poole, JH and Byrne, RW 2007, 'Elephants classify human

ethnic groups by odor and garment colour', Current Biology, vol. 17, no. 22,

doi:10.1016/j.cub.2007.09.060.

Burn, RW, Underwood, FM & Blanc, J 2011, 'Global trends and factors associated with the illegal killing

of elephants: a hierarchical Bayesian analysis of carcass encounter data', PLoS ONE, vol. 6, no. 9,

e24165. doi:10.1371/journal.pone.0024165.

Campos-Arceiz, A, Takatsuki, S, Ekanayaka, SKK & Hasegawa, T 2009, 'The human-elephant conflict in

southeastern Sri Lanka: type of damage, seasonal patterns, and sexual differences in the raiding

behaviour of elephants', Gajah, vol. 31, pp. 5 – 14.

Chakraborty, K & Mondal, J 2013, 'Perceptions and patterns of human-elephant conflict at Barjora

block of Bankura district in West Bengal, India: insights for mitigation and management',

Environmental Development and Sustainability, vol. 15, pp. 547 – 565.

Changa, A, de Souza, N, Muya, J, Keyyu, J, Mwakatobe, A, Malugu, L, Ndossi, HP, Konuche, J, Omondi,

R, Mpinge, A, Hahn, N, Palminteri, S and Olson, D 2016, 'Scaling-up the use of chilli fences for reducing

human-elephant conflict across landscapes in Tanzania', Tropical Conservation Science, vol. 9, no. 2,

pp. 921 – 930.

Chiyo, PI, Cochrane, EP, Naughton, L & Basuta, GI 2005, 'Temporal patterns of crop raiding by

elephants: a response to changes in forage quality or crop availability?', African Journal of Ecology,

vol. 43, pp. 48 – 55.

Choudhury, A 2004, 'Human-elephant conflicts in northeast India', Human Dimensions of Wildlife, vol.

9, no. 4, pp. 261 – 270.

Cook, RM, Witkowski, ETF, Helm, CV, Henley, MD & Parrini, F 2017, 'Recent exposure to African

elephants after a century of exclusion: rapid accumulation of marula tree impact and mortality, and

poor regeneration', Forest Ecology and Management, vol. 401, pp. 107 – 116.

18

Das, BJ, Saikia, BN, Baruah, KK, Bora, A & Bora, M 2014, 'Nutritional evaluation of fodder, its preference

and crop-raiding by wild Asian elephant (Elephas maximus) in Sonitpur District of Assam, India',

Veterinary World, vol. 7(12), pp. 1082-1089.

Davies, TE, Wilson, S, Hazarika, N, Chakrabaty, J, Das, D, Hodgson, DJ & Zimmerman, A

2011,'Effectiveness of intervention methods against crop-raiding elephants', Conservation Letters,

doi: 10.1111/j.1755-263X.2011.00182.x

de Silva, M 1998, 'Status and conservation of the elephant (Elephas maximus) and the alleviation of

man-elephant conflict in Sri Lanka', Gajah, vol. 19, pp. 1 – 68.

de Silva, S, Webber, CE, Weerathugna, US, Pushpakumara, TV, Weerakoon, DK and Wittemyer, G 2013,

'Demographic variables for wild Asian elephants using longitudinal observations', PLoS ONE, vol. 8, no.

12, e82788. doi: 10.1371/journal.pone.0082788.

Dickman, AJ 2010, 'Complexities of conflict: the importance of considering social factors for effectively

resolving human-wildlife conflict', Animal Conservation, vol. 13, pp. 458 - 466.

Dublin, HT & Hoare, RE 2004, 'Searching for solutions: the evolution of an integrated approach to

understanding and mitigating human-elephant conflict in Africa', Human Dimensions of Wildlife, vol.

9, no. 4, pp. 271 – 278.

Dutta, A 2013, 'Bees, acrid smoke from chillis keep elephants away and reduce conflict', The Times of

India, June 24, 2013.

Ekanayaka, SKK, Campos-Arceiz, A, Rupasinghe, M, Pastorini, J & Fernando, P 2011, 'Patterns of crop

raiding by Asian elephants in a human-dominated landscape in southeastern Sri Lanka', Gajah, vol. 34,

pp. 20 -25.

Evans, LA and Adams, WM 2018, 'Elephants as actors in the political ecology of human-elephant

conflict', Transactions of the Institute of British Geographers, doi: 10.1111/tran.12242.

19

Fernando, P, de Silva, MKCR, Jayasinghe, LKA, Janaka, HK & Pastorini, J 2019, ‘First country-wide survey

of the endangered Asian elephant: toward better conservation and management in Sri Lanka’, Oryx,

pp. 1 – 10, https://doi.org/10.1017/s0030605318001254 .

Fernando, P, Jayewardene, J, Prasad, T, Hendavitharana, W & Pastorini, J 2011, ‘Current status of Asian

Elephants in Sri Lanka’, Gajah, vol. 25, pp. 93 – 103.

Fernando, P, Kumar, MA, Williams, AC, Wikramanayake, E, Aziz, T & Singh, SM 2008, Review of human-

elephant conflict mitigation measures practiced in South Asia, AREAS Technical Support Document,

World Wide Fund for Nature.

Fernando, P, Wikramanayake, E, Weerakoon, D, Jayasinghe, LKA, Gunawardene, M & Janaka, HK 2005,

'Perceptions and patterns of human-elephant conflict in old and new settlements in Sri Lanka: insights

for conservation and management', Biodiversity and Conservation, vol. 14, pp. 2465 – 2481.

Fischhoff, IR, Sundaresan, SR, Cordingley, J and Rubenstein, DI 2007, 'Habitat use and movements of

plains zebra (Equus burchelli) in response to predation danger from lions', Behavioural Ecology, doi:

10.1093/beheco/arm036.

Freeman, HD, Brosnan, SF, Hopper, LM, Lambeth, SP, Schapiro, SJ & Gosling, SD 2013, 'Developing a

comprehensive and comparative questionnaire for measuring personality in chimpanzees using a

simultaneous top-down/bottom-up design', American Journal of Primatology, vol. 75, pp. 1042 –

1053.

Gillingham, S & Lee, PC 2003, 'People and protected areas: a study of local perceptions of wildlife crop-

damage conflict in an area bordering the Selous Game Reserve, Tanzania', Oryx, vol. 37, pp. 316 – 325.

Graham, MD, Notter, B, Adams, WM, Lee, PC & Ochieng, TN 2010, 'Patterns of crop-raiding by

elephants, Loxodonta africana, in Laikipia, Kenya, and the management of human-elephant conflict',

Systematics and Biodiversity, vol. 8, no. 4, pp. 435 – 445.

Graham, MD, Douglas-Hamilton, I, Adams, WM, and Lee, PC 2009, 'The movement of African

elephants in a human-dominated land-use mosaic', Animal Conservation, vol. 12, pp. 445 – 455.

20

Graham, MD & Ochieng, T 2008, 'Uptake and performance of farm-based measures for reducing crop-

raiding by elephants Loxodonta Africana among smallholder farms in Laikipia District, Kenya', Oryx,

vol. 42, pp. 76 – 82.

Gross, EM, Drouet-Hoguet, N, Subedi, N & Gross, J 2017, 'The potential of medicinal and aromatic

plants (MAPs) to reduce crop damages by Asian elephants (Elephas maximus)', Crop Protection, vol.

100, pp. 29 – 37.

Gross, EM, Mc Robb, R and Gross, J 2016, 'Cultivating alternative crops reduces crop losses due to

African elephants' Journal of Pest Science, vol. 89, no. 2, pp. 495 – 506.

Gubbi, S 2012, 'Patterns and correlates of human-elephant conflict around a south Indian reserve',

Biological Conservation, vol. 148, pp. 88 – 95.

Gunaryadi, D, Sugiyo, Hedges, S 2017, 'Community-based human-elephant conflict mitigation: the

value of an evidence-based approach in promoting the uptake of effective methods', PLoS ONE, vol.

12, no. 5, e0173742.

Harich, FK, Treydte, AC, Sauerborn, J & Owusu, EH 2013, 'People and wildlife: conflicts arising around

the Bia Conservation Area in Ghana', Journal for Nature Conservation, vol. 21, pp. 342 – 349.

Hartter, J, Goldman, A & Southworth, J 2011, 'Responses by households to resource scarcity and

human-wildlife conflict: issues of fortress conservation and the surrounding agricultural landscape',

Journal for Nature Conservation, vol. 19, pp. 79 – 86.

Haturusinghe, HS & Weerakoon, DK 2012, ‘Crop raiding behaviour of elephants in the Northwestern

region of Sri Lanka’, Gajah, vol. 36, pp. 26 – 31.

Hedges, S & Gunaryadi, D 2009, 'Reducing human-elephant conflict: do chillies help deter elephants

from entering crop fields?', Oryx, vol. 44, no. 1, pp. 139 – 146.

Hennessey, DJG n.d, 'I watch in a chena', IN: (eds: Ranasinghe, D, Ratnapala, S, Seneviratne, L,


Volume 1 (1936) to Volume 10 (1964), Nedimala, YN Print Graphics, pp. 51 – 55.

21

Hennessey, DJG 1929, 'Elephant shikar', IN: (eds: Ranasinghe, D, Ratnapala, S, Seneviratne, L,


Volume 1 (1936) to Volume 10 (1964), Nedimala, YN Print Graphics, pp. 9 -13.

Hill, CM 2004, 'Farmers' perspectives of conflict at the wildlife-agricultural boundary: some lessons

learned from African subsistence farmers', Human Dimensions of Wildlife, vol. 9, pp. 279 – 286.

Hill, CM 2000, 'Conflict of interest between people and baboons: crop raiding in Uganda', International

Journal of Primatology, vol. 21, no. 2, pp. 299 – 315.

Hoare, R 2015, 'Lessons from 20 years of human-elephant conflict mitigation in Africa', Human

Dimensions of Wildlife, vol. 20, pp. 289 – 295.

Hoare, R 2000, 'African elephants and humans in conflict: the outlook for co-existence', Oryx, vol. 34,

pp. 34 – 38.

Hoffmeier-Karimi, RR & Schulte, BA 2014, 'Assessing perceived and documented crop damage in a

Tanzanian village impacted by human-elephant conflict (HEC)', Pachyderm, vol. 56, pp. 51 – 60.

Ihwagi, F, Thouless, C, Wang, T, Skidmore, AJ, Omondi, P & Douglas-Hamilton, I 2018, 'Night-day speed

ratio of elephants as indicator of poaching levels', Ecological Indicators, vol. 84, pp. 38 – 44.

Kahindi, O, Wittemyer, G, King, J, Ihwagi, F, Omondi, P & Douglas-Hamilton, I 2009, 'Employing

participatory surveys to monitor the illegal killing of elephants across diverse land uses in Laikipia-

Samburu, Kenya', African Journal of Ecology, doi: 10.1111/j.1365-2028.2009.01.200.x.

Karidozo, M & Osborn, FV 2015, 'Community based conflict mitigation trials: results of field tests of

chilli as an elephant deterrent', Biodiversity and Endangered Species, vol. 3 no. 1,

http://dx.doi.org/10.4172/2332-2543.1000144.

King, L, Pardo, M, Weerathunga, S, Kumara, TV, Jayasena, N, Soltis, J & de Silva, S 2018, ‘Wild Sri Lankan

elephants retreat from the sound of disturbed Asian honey bees’, Current Biology, vol. 28, pp 64 – 65.

22

King, LE, Lala, F, Nzumu, H, Mwambingu, E & Douglas-Hamilton, I 2017, 'Beehive fences as a

multidimensional conflict-mitigation tool for farmers coexisting with elephants', Conservation Practice

and Policy, vol. 31, no. 4, pp. 743 – 752.

King, LE, Douglas-Hamilton, I & Vollrath, F 2011, 'Beehive fences as effective deterrents for crop-

raiding elephants: field trials in northern Kenya', African Journal of Ecology, doi: 10.1111/j.1365-

2028.2011.01275.x.

King, LE, Soltis, J, Douglas-Hamilton, I, Savage, A & Vollrath, F 2010, 'Bee threat elicits alarm call in

African elephants', PLoS ONE, vol. 5, no. 4, e10346. doi:10.1371/journal.pone0010346.

King, LE, Lawrence, A, Douglas-Hamilton, I & Vollrath, F 2009, 'Beehive fence deters crop-raiding

elephants', African Journal of Ecology, vol. 47, pp. 131 – 137.

King, LE, Douglas-Hamilton, I & Vollrath, F 2007, 'African elephants run from the sounds of disturbed

bees', Current Biology, vol. 17, no. 19, R832.

Lacy, KE & Martins, EP 2003, 'The effect of anthropogenic habitat usage on the social behaviour of a

vulnerable species, Cyclura nubila', Animal Conservation, vol. 6, pp. 3 – 9.

Lewis, DL, Baruch-Mordo, S, Wilson, KR, Breck, SW, Mao, JS & Broderick, J 2015, 'Foraging ecology of

black bears in urban environments: guidance for human-bear conflict mitigation', Ecosphere, vol. 6,

no. 8, http://dx.doi.org/10.1890/ES15-00137.1.

Loe, J & Roskaft, E 2004, 'Large carnivores and human safety: a review', Ambio, vol. 6, pp. 283 – 288.

Lord, A, Waas, JR & Innes, J 1997, 'Effects of human activity on the behaviour of northern New Zealand

dotterel Charadrius obscurus aquilonius chicks', Biological Conservation, vol. 82, pp. 15 – 20.

Madden, F 2004, 'Creating coexistence between humans and wildlife: global perspectives on local

efforts to address human-wildlife conflict', Human Dimensions of Wildlife, vol. 9, no. 4, pp. 247 – 257.

Manfredo, MJ 2015, 'Essays on human-wildlife conflict 10 years after the Durban World Parks

Congress: an introduction', Human Dimensions of Wildlife, vol. 20, pp. 285 – 288.

23

Manoj, EM 2012, 'Jumbos get a sting of defeat', The Hindu Newspaper, February 25, 2012.

Meixner, MD and Conte, YL 2016, 'A current perspective on honey bee health', Apidologie, vol. 47, pp.

273 – 275.

Menard, N, Foulquier, A, Vallet, D, Qarro, M, Le Gouar, P & Pierre, JS 2013, 'How tourism and

pastoralism influence population demographic changes in a threatened large mammal species',

Animal Conservation, vol. 17, pp. 115-124.

Nair, RP and Jayson, EA 2016, 'Effectiveness of beehive fences to deter crop raiding elephants in

Kerala, India', International Research Journal of Natural and Applied Sciences, vol. 3, no. 1, pp. 14 – 19.

Naughton-Treves, L & Treves, A 2005, 'Socio-ecological factors shaping local support for wildlife: crop-

raiding by elephants and other wildlife in Africa', IN: (eds: Woodroffe R, Thirgood S and Rabinowitz A)

People and wildlife: conflict or coexistence? Cambridge, Cambridge University Press, pp. 252 – 277.

Nelson, A, Bidwell, P and Sillero-Zubiri 2003, 'A review of human-elephant conflict mitigation

strategies', People and Wildlife Initiative, Wildlife Conservation Research Unit, Oxford University.

Ngama, S, Korte, L, Bindelle, J, Vermeulen, C & Poulsen, JR 2016, 'How bees deter elephants: beehive

trials with forest elephants (Loxodonta africana cyclotis) in Gabon', PLoS ONE, vol. 11, no. 5, e0155690.

doi:10.1371/journal.pone.0155690.

Nijman, V & Nekaris, KA 2010, 'Testing a model for predicting primate crop-raiding using crop-and-

farm specific risk values', Applied Animal Behaviour Science, vol. 127, pp. 125 – 129.

Ogra, MV 2008, 'Human-wildlife conflict and gender in protected area borderlands: a case study of

costs, perceptions and vulnerabilities from Uttarakhand (Uttaranchal), India', Geoforum, vol. 39, pp.

1408 – 1422.

Okello, MM, Kiringe, JW, Njumbi, SJ and Isiche, J 2016, 'Prevalence of human-elephant conflicts in

Amboseli ecosystem, Kenya: current opinions of local communiy', International Journal of Biodiversity

and Conservation, vol. 8, no. 3, pp. 60 – 71.

24

Olson, TL, Gilbert, BK & Squibb, RC 1997, 'The effects of increasing human activity on brown bear use

of an Alaskan river', Biological Conservation, vol. 82, pp. 95 – 99.

Osborn, FV & Parker, GE 2003, 'Towards an integrated approach for reducing the conflict between

elephants and people: a review of current research', Oryx, vol. 37, no. 1, doi.

10.1017/90030605302000000.

Perera, BMAO 2009, 'The human-elephant conflict: a review of current status and mitigation

methods', Gajah, vol. 1, pp. 341 – 344.

Perez, E & Pacheco, LF 2006, 'Damage by large mammals to subsistence crops within a protected area

in a montane forest of Bolivia', Crop Protection, vol. 25, pp. 933 – 939.

Ponnusamy, V, Chackrapani, P, Lim, TW, Saaban, S and Campos-Arceiz, A 2016, 'Farmers' perceptions

and attitudes towards government constructed electric fences in Peninsular Malaysia', Gajah, vol. 45,

pp. 4 – 11.

Pozo, RA, Coulson, T, McCulloch, G, Strongza, A & Songhurst, A 2017, 'Chilli-briquettes modify the

temporal behaviours of elephants, but not their numbers', Oryx, doi: 10.1017/S0030605317001235.

Punchihewa, RWK 1994, Beekeeping for honey production, Sarvodaya Visha Lekha Publishing, Sri

Lanka.

Redpath, SM, Young, J, Evely, A, Adams, WM, Sutherland, WJ, Whitehouse, A, Amar, A, Lambert, RA,

Linnell, JDC, Watt, A & Gutierrez, RJ 2013, 'Understanding and managing conservation conflicts',

Trends in Ecology and Evolution, vol. 28, no. 2, pp. 100 – 109.

Rao, KS, Maikhuri, RK, Nautiyal, S & Saxena, KG 2002, 'Crop damage and livestock depredation by

wildlife: a case study from Nanda Devi Biosphere Reserve, India', Journal of Environmental

Management, vol. 66, pp. 317 – 327.

Ripple, WJ, Abernethy, K, Betts, MG, Chapron, G, Dirzo, R, Galetti, M, Levi, T, Lindsey, PA, Macdonald,

DW, Machovina, B, Newsome, TM, Peres, CA, Wallach, AD, Wolf, C & Young, H 2016, 'Bushmeat

25

hunting and extinction risk to the world's mammals', Royal Society of Open Science, vol. 3, pp. 160498.

http://dx.doi.org/10.1098/rsos.160498.

Saj, T, Sicotte, P & Paterson, JD 1999, 'Influence of food consumption on the time budget of vervets',

International Journal of Primatology, vol. 20, pp. 977 – 994.

Santiapillai, C, Wijeyamohan, S, Bandara, G, Athurupana, R, Dissanayake, N & Read, B 2010, 'An

assessment of the human-elephant conflict in Sri Lanka', Ceylon Journal of Biological Sciences, vol. 39,

no. 1, pp. 21 – 33.

Santiapillai, C, Fernando, P and Gunewardene, M 2006, 'A strategy for the conservation of the Asian

elephant in Sri Lanka', Gajah, vol. 25, pp. 91 – 102.

Santiapillai, C 1996, 'Mitigation of human-elephant conflicts in Sri Lanka', Gajah, vol. 15, pp. 1 – 7.

Santiapillai, C & de Silva, M 1994, 'An action plan for the conservation and management of elephant

(Elephas maximus) in Sri Lanka', Gajah, vol. 13, pp. 1 – 24.

Saraswat, R, SInha, A & Radhakrishna, S 2015, 'A god becomes a pest? Human-rhesus macaque

interactions in Himachal Pradesh, northern India', European Journal of Wildlife Research, vol. 61, pp.

435 – 443.

Scheijen, CPJ, Richards, SA, Smit, J, Jones, T & Nowak, K 2018, 'Efficacy of beehive fences as barriers

to African elephants: a case study in Tanzania', Oryx, doi: 10.1017/S0030605317001727.

Sitati, NW & Walpole, MJ 2006, 'Assessing farm-based measures for mitigation of human-elephant

conflict in Transmara District, Kenya', Oryx, vol. 40, no. 3, pp. 279 – 286.

Sitati, NW, Walpole, MJ, Smith, RJ & Leader-Williams, N 2003, 'Predicting spatial aspects of human-

elephant conflict', Journal of Applied Ecology, vol. 40, pp. 667 – 677.

Srinivasaiah, NM, Anand, VD, Vaidyanathan, S & Sinha, A 2012, 'Usual populations, unusual

individuals: insights into the behaviour and management of Asian elephants in fragmented

landscapes', PLoS ONE, vol. 7, no. 8, e42571. doi: 10.1371/journal.pone.0042571.

26

Studsrod, JE & Wegge, P 1995, 'Park-people relationships: the case of damage caused by park animals

around the Royal Bardia National Park, Nepal', Environmental Conservation, vol. 22, no. 2, pp. 132 –

142.

Sukumar, R 2006, 'A brief review of the status, distribution and biology of wild Asian elephants Elephas

maximus', International Zoo Yearbook, vol. 40, pp. 1 – 8.

Sukumar, R 1990, 'Ecology of the Asian elephant in southern India. II. Feeding habits and crop raiding

patterns', Journal of Tropical Ecology, vol. 6, pp. 33 – 53.

Thapa, S 2010, 'Effectiveness of crop protection methods against wildlife damage: a case study of two

villages at Bardia National Park, Nepal', Crop Protection, vol. 29, pp. 1297 – 1304.

Thaufeek, UL, Padmalal, UKGK and Fernando, P 2014, 'Land use and human-elephant conflict in the

Sigiriya Sanctuary, Sri Lanka', Gajah, vol. 40, pp. 26 – 30.

The Guardian, 2011, 'Sri Lanka elephant survey puts population at 5,879', The Guardian (online) 2

September, https://www.theguardian.com/world/2011/sep/02/sri-lanka-elephant-survey-

population (Accessed 24 September 2018).

Thirgood, S, Woodroffe, R and Rabinowitz, A 2005, 'The impact of human-wildlife conflict on human

lives and livelihoods', IN: (eds: Woodroffe R, Thirgood S and Rabinowitz A) People and wildlife: conflict

or coexistence? Cambridge, Cambridge University Press, pp. 13 – 26.

Thirunavukarasu, V 2014, Assessment of human elephant conflicts and mitigation measures to resolve

the conflict in Dharamapuri Forest Division, Tamil Nadu, India, MCT Phase IV Assignment, District

Forest Office, Tamil Nadu, India.

Thuppil, V and Coss RG 2013, 'Wild Asian elephants distinguish aggressive tiger and leopard growls

according to perceived danger', Biology Letters, Vol. 9, 20130518.

http://dx.doi.org/10.1098/rsbl.2013.0518.

https://www.theguardian.com/world/2011/sep/02/sri-lanka-elephant-survey-population

https://www.theguardian.com/world/2011/sep/02/sri-lanka-elephant-survey-population

http://dx.doi.org/10.1098/rsbl.2013.0518

27

Tutein-Nolethnius, AC n.d. 'Is our emblem doomed?' IN: (eds: Ranasinghe, D, Ratnapala, S,

Seneviratne, L, Schwammer, H, Schwammer, G and Perera, BH) Saga of the elephant in Sri Lanka:

Extracts from Loris Volume 1 (1936) to Volume 10 (1964), Nedimala, YN Print Graphics, pp. 20 – 21.

van de Water, A and Matteson, K 2018, 'Human-elephant conflict in western Thailand: socio-economic

drivers and potential mitigation strategies', PLoS ONE, vol. 13, no. 6, e0194736.

Van Eden, M, Ellis, E and Bruyere, BL 2016, 'The influence of human-elephant conflict on electric fence

management and perception among different rural communities in Laikipia County, Kenya', Human

Dimensions of Wildlife, vol. 21, no. 4, pp. 283 – 296.

Vollrath, F & Douglas-Hamilton, I 2002, 'African bees to control African elephants',

Naturwissenschaften, vol. 89, pp. 508 – 511.

von Gerhardt, K, Niekerk, AV, Kidd, M, Samways, M & Hanks, J 2014, 'The role of elephant Loxodonta

africana pathways as a spatial variable in crop-raiding location', Oryx, vol. 48(3), pp. 436 – 444.

Webber, CE, Sereivathana, T, Maltby, MP and Lee, PC 2011, 'Elephant crop-raiding and human-

elephant conflict in Cambodia: crop selection and seasonal timing of raids', Oryx, vol. 45, no. 2, pp.

243 – 251.

Weerakoon, DK, Gunawardene, MD, Janaka, HK, Jayasinghe, LKA, Perera, RAR, Fernando, P and

Wikramanayake, E 2004, 'Ranging behaviour and habitat use of elephants in Sri Lanka', in: (ed)

Jayewardene, J Endangered Elephants: Past, present and future, Proceedings of the symposium for

human-elephant relationships and conflicts, Sri Lanka, September 2003. Biodiversity and Conservation

Trust, Colombo, pp. 68 - 70.

Wikramanayake, ED, Hathurusinghe, HS, Janaka, HK, Jayasinghe, LKA, Fernando, P, Weerakoon, DK &

Gunawardene, MD 2004, 'The human-elephant conflict in Sri Lanka: lessons for mitigation,

management, and conservation from traditional land-use patterns', in: (ed) Jayewardene, J

Endangered Elephants: Past, present and future, Proceedings of the symposium for human-elephant

relationships and conflicts, Sri Lanka, September 2003. Biodiversity and Conservation Trust, Colombo,

pp. 164-169.

28

Williams, AC, Johnsingh, AJT & Krausman, PR 2001, 'Elephant-human conflicts in Rajaji National Park,

north-western India', Wildlife Society Bulletin, vol. 29, no. 4, pp. 1097 – 1104.

Wilson, ML, Hauser, MD & Wrangham, RW 2007, 'Chimpanzees (Pan troglodytes) modify grouping and

vocal behaviour in response to location-specific risk', Behaviour, vol. 144, pp. 1621 – 1653.

Winston, ML 1987, The biology of the honey bee, Harvard University Press 2nd Edition, USA, pp. 4 – 12.

Woodroffe, R, Lindsey, PA, Romanach, SS, Stein, A, Ranah, SMKO 2005a, 'Livestock predation by

endangered African wild dogs (Lycaon pictus) in northern Kenya', Biological Conservation, vol. 124, pp.

225 – 234.

Woodroffe, R, Thirgood, S & Rabinowitz, A (Eds) 2005b, People and wildlife: conflict or coexistence?

Cambridge University Press, Cambridge.

29

Chapter 2: General behaviour patterns of wild elephants utilising areas of varying anthropogenic risk to elephants, in Sri Lanka

30

2.1 Abstract

Many species alter their behaviour in response to perceived levels of anthropogenic disturbance, or

risk, in a given environment. Risk-avoidance responses such as fleeing are typical, and often have

fitness-consequences as the time available for important activities such as feeding, mating or resting

is reduced. In this study, the behaviour of Asian elephants observed in areas representing a low or

medium level of anthropogenic risk was investigated. The results provided no evidence that general

behavioural patterns including time spent in locomotion, feeding or standing still differed between

elephants observed inside National Park boundaries (low-risk zone) or in nearby external forest

(medium-risk zone). The frequency with which elephants performed typical ‘reactive’ behaviours,

such as smelling or dusting, was also similar. However, when presented with an immediate

anthropogenic disturbance, such as an approaching vehicle, elephants would reduce the time spent

feeding and increase the frequency of ‘reactive’ behaviours regardless of whether they were observed

in the low- or medium risk zone. Overall, this study provides initial evidence that certain human-

influenced stimuli are perceived as a disturbance or threat by elephants, while also highlighting the

challenging nature of investigating elephant response to anthropogenic disturbance in the wild.

2.2 Introduction

A by-product of worldwide human population growth and the correlating infrastructure and

agricultural development is a blurring of the delineation between human and wildlife habitats

(Fernando et al. 2011; Graham et al. 2009). Inevitably, this results in an increase of interactions

between the species, which is influencing a habitat of conflict between human and wildlife

communities (Lacy and Martins 2003). Anthropogenically induced alterations reducing the size and

quality of wildlife habitat, drive competition for resources both within and between species and force

many non-human animals to adjust to unprecedented levels of human activity (Lacy and Martins

2003). Termed ‘anthropogenic disturbance’ these interactions challenge the health and long-term

viability of numerous wildlife species (Setsaas et al. 2007; Frid and Dill 2002; Lima 1998).

Anthropogenic disturbance often leads to situations or environments of perceived or actual riskiness

to animals (Lacy and Martins 2003; Frid and Dill 2002; Anthony and Blumstein 2000; Mattson et al.

1992), and many species alter their behaviour to attend to varying risks in their environment (Lacy and

31

Martins 2003; Frid and Dill 2002). For species which are hunted, captured or otherwise injured by

humans, the risks associated with anthropogenic disturbance are particularly acute (Caro 2005; Frid

and Dill 2002). In such situations, animals often respond in a manner akin to responses provoked by

the threat of predation, of which risk-avoidance or minimisation strategies are common (Frid and Dill

2002; Steidl and Anthony 2000). However, the obvious benefits of increased safety assumed by risk-

avoidance, often incur costs to fitness including less time available for feeding, resting or mating

(Blumstein et al. 2005; Lacy and Martins 2003; Frid and Dill 2002; Anthony and Blumstein 2000; Lima

1998), excess energy expended while guarding or fleeing (Kiffner et al. 2014; Frid and Dill 2002), or

reduced access to optimal habitat (Blumstein et al. 2005; Frid and Dill 2002; de la Torre et al. 2000;

Lima 1998; Lima and Dill 1990). Such compromises to vital activities can have both short- and long-

term negative consequences on body condition, reproductive success and even survival for individual

animals (White et al. 2002; Lima 1998; Lima and Dill 1990).

Identifying the ways in which wild animals respond to anthropogenic disturbance is fraught with

difficulty. Behavioural change can often happen slowly, and by the time it is suspected it is often too

late to acquire a baseline of formerly typical behaviour by which to compare. Researchers have

addressed this challenge by comparing the behaviour of wildlife thought to be, or at risk of being,

affected by anthropogenic disturbance at locations representing varying levels of human activity. This

can include comparisons of areas of low to high hunting pressure (e.g. de Boer et al. 2004), areas

representing varying levels of non-lethal human recreational activity (e.g. Lusseau et al. 2009; White

et al. 1999), or inside and outside of designated wildlife protected areas (PA’s) (Crosmary et al. 2012).

Such studies provide numerous examples of behavioural differences according to perceived level of

anthropogenic risk, across a wide array of species (e.g. Crosmary et al. 2012; Lusseau et al. 2009;

Griffin et al. 2007; Setsaas et al. 2007; Donadio and Buskirk 2006; Williams et al. 2006; Caro 2005;

Lusseau 2003; White et al. 1999).

For example, recent studies show that heavily exploited mammals in Tanzania (including buffalo,

giraffe, roan antelope, topi, waterbuck, warthog and zebra) are more likely to flee or increase vigilance

in the presence of human observers outside of PA’s (Caro 2005). Similar avoidance behaviours were

exhibited in marine mammals who modified diving patterns to spend more time deeper underwater,

in response to the presence of tourist vessels (e.g. dolphins: Lusseau 2003; and killer whales: Lusseau

et al. 2009; Williams et al. 2006). For dolphins specifically, this change in locomotive patterns

32

interrupts important foraging opportunities (Lusseau 2003). Similarly, in the United States of America

(USA), grizzly bears increase time spent moving, thus reducing available foraging time, in areas where

human recreational activities such as mountain climbing are popular (White et al. 1999). This inverse

relationship between movement and feeding is also exhibited in camelid species inhabiting reserves

in Argentina where poaching frequently occurs (Donadio and Buskirk 2006). Other species, while not

necessarily altering locomotive patterns, exhibit increased vigilance behaviours, which can also

interrupt foraging. Heightened vigilance levels have been observed in roe deer occupying areas of

frequent hunting in the Netherlands (de Boer et al. 2004), in several species of ungulates utilising

habitat outside of PA’s in Zimbabwe (Crosmary et al. 2012) and Tanzania (Setsaas et al. 2007), and by

Olympic marmots in areas of high human recreation activity in the USA (Griffin et al. 2007).

As highlighted above, many species demonstrate a capacity to recognise variations in risk between

areas representing different levels of anthropogenic disturbance. Additionally, many species can also

differentiate between fluctuations in anthropogenic risk within a given location (Ferrari et al. 2009;

Bates et al. 2007), particularly if the risk is heightened by the physical presence of humans rather than

just anthropogenically induced changes to the ecosystem. Fluctuations in risk can occur daily (e.g. the

physical presence of a predator, or increasing visitor proximity to self), or seasonally (e.g. consistent

increased human presence during peak tourist holidays or legal hunting seasons) (Lima and Dill 1990).

An ability to discriminate between risk levels enables an animal to respond appropriately to the

specific situation, rather than expending excess energy in a constant state of high alert (Ferrari et al.

2009; Bates et al. 2007; Lima and Bednekoff 1999). For example, European roe deer increase vigilance

behaviours during assigned hunting seasons, then relax these behaviours when hunting ceases

(Behhaiem et al. 2008). Similarly, Alaskan brown bears who were not yet habituated to human

presence, would delay salmon feeding if the tourist season was extended, despite high availability of

food (Olson et al. 1997).

While changes in locomotion, foraging and vigilance are common responses to anthropogenic

disturbance, they are not necessarily uniform within individuals of the species present, and sex-biased

differences in response are quite common (Bunnerfeld et al. 2006; Lusseau 2003; Saj et al. 1999).

Generally speaking, males are likely to engage in riskier behaviour than females, trading off costs of

potential capture or injury for rewards including greater quantities of forage, or more nutritionally

rich foods (Ekananyaka et al. 2011; Bunnerfeld et al. 2006; Saj et al. 1999, Sukumar and Gadgil 1988).

33

This was evident in the Eurasian lynx, where male lynx spent more time in proximity to human

habitations than did female lynx, presumably driven by the increased density of prey species in these

locations and more willing to accept the associated anthropogenic risks (Bunnerfeld et al. 2006). A

similar story presents with male vervet monkeys who actively seek out anthropogenically disturbed

areas to forage on crops far more frequently than females (Saj et al. 1999). However, there are

exceptions. For example, while dolphins of both sexes interrupted foraging patterns and altered

locomotion by deep-diving in response to approaching tourist vessels, males were the most risk-

averse and initiated diving earlier than females (Lusseau 2003).

Elephants, both African and Asian, exemplify species whose long-term survival and wellbeing is

threatened by increasing levels of anthropogenic disturbance (Graham et al. 2010; Perera 2009; Sitati

and Walpole 2006; Bandara and Tisdell 2002). Elephant habitat in all elephant range countries has

been severely depleted due to human driven development and encroachment forcing humans and

elephants into ever-increasing proximity (Amwata and Mganga 2014; Chakraborty and Mondal 2013;

Campos-Arceiz et al. 2009; Fernando et al. 2005; Choudhury 2004; Sitati et al. 2003). Aside from a rare

few records of lions or tigers preying on young or weak elephants (lions: Davidson et al. 2013; Power

and Compion 2009; Loveridge et al. 2006, tigers: Kumaraguru et al. 2011), humans are the only

‘predator’ of elephants. Historically, humans hunted elephants for ‘sport’ and population control, and

captured elephants for entertainment, warfare and labour (Rangarajan 1996; Sukumar 1989).

Presently, poaching for body parts (Ihwagi et al. 2018; Burn et al. 2011; Kahindi et al. 2009) and

retaliatory attacks by farmers, who are frustrated and irate following elephant raids on their crops,

remain major threats to elephants (Okello 2016; Chakraborty and Mondal 2013; Gubbi 2012; Davies

et al. 2011; Webber et al. 2011; Graham et al. 2010; Perera 2009; Sitati et al. 2003). Thus, it makes

sense that elephants perceive humans as a threat (Thuppil and Goss 2013). Understanding how these

long-lived and cognitively sophisticated animals adapt behaviours to negotiate a world where their

habitat is increasingly altered by anthropogenic disturbance, could be key to developing effective

management strategies and promoting a more harmonious co-existence between humans and

elephants.

As it stands, evidence exists from multiple locations supporting the notion that elephants are aware

of the risks of anthropogenic disturbance and activity, and that behavioural adjustments are made to

minimise the risks presented by interactions with humans (Wilkie and Douglas-Hamilton 2019; Gunn

34

et al. 2013; Graham et al. 2009; Barnes et al. 2006). In rural communities across both Africa and Asia

where crop-raiding by elephants is common, elephants forage on crops almost exclusively at night

(Gunn et al. 2013; Graham et al. 2009; Barnes et al. 2006), despite wild foraging occurring often during

the day (Sukumar 1989). African elephants have also been known to increase their average speed of

locomotion when traversing areas of high human density (Wilkie and Douglas-Hamilton 2019).

Furthermore, in Kenya, elephants responded to the perceived presence of humans from specific tribes

known to present a threat, with increased vigilance and locomotion when compared to other less

harmful stimuli (Soltis et al. 2014; Bates et al. 2007). However, these examples typically focus on the

upper and lower scales of risk, comparing behaviour in the riskiest situations (e.g. foraging on crops in

a human occupied village) and the least risky (e.g. within the confines of a PA). More and more,

elephants are spending increased time outside of PA’s, especially in Asia where electric fences are less

common and often poorly maintained (Ponnusamy et al. 2016; Fernando et al. 2011; Santiapillai et al.

2006). The majority of elephants in Sri Lanka spend at least some of their time outside of PA’s, which

is not surprising given that more than half of suitable elephant habitat falls outside of protected

boundaries (Fernando et al. 2019; Santiapillai et al. 2006; Weerakoon et al. 2004). This increase in

proximity to humans certainly poses additional risks and exposure to anthropogenic disturbance for

elephants however is not as extreme as when elephants come into direct conflict with humans, such

as during crop raids. Identifying whether elephants adapt behaviours in this interim environment of

risk, and if so how, would provide clues to a largely unexplored element of human-elephant

coexistence.

When investigating and comparing behavioural patterns of any wild animal across multiple sites, it is

impossible to account for all confounding variables. This is, of course, the situation when studying wild

elephant behaviour in areas representing different levels of anthropogenic disturbance. A range of

factors, from even subtle environmental variations (e.g. vegetative structure, availability of resources,

microclimates, presence or absence of other species) (Griffin 2007) to individual animal personalities

(Hollander et al. 2008; Freeman et al. 2007) can potentially influence behaviour. However, logistic and

ethical constraints render it nigh on impossible to experimentally manipulate anthropogenic

disturbance in the wild. Captive studies can be useful but come with their own practical and ethical

considerations, and comparability may be debatable particularly in relation to long-lived and

intelligent species. That studies spanning numerous continents and species have demonstrated

35

changes in behaviour consistent with the scale of anthropogenic disturbance presented at different

locations, shows merit in such comparative approaches despite some limitations.

In this study, we provide a comparison of general behaviour patterns, specifically locomotion, foraging

and standing still, for Asian elephants utilising habitat representing different levels of anthropogenic

disturbance or risk to elephants. We selected an area of ‘low’ anthropogenic disturbance (hereafter

‘low risk zone’) located within designated wildlife PA boundaries, and an area of ‘medium’

anthropogenic disturbance (hereafter ‘medium risk zone’), located outside of the PA. Specifically, we

aimed to:

- Compare the time elephants spent foraging, in locomotion, and ‘standing still without feeding’

in the low and medium risk zones

- Compare frequency of ‘reactive behaviours’ (backing away, dusting, smelling, ‘tail up’ posture,

throwing objects, trunk touch to own mouth, vigilance/freeze posture) observed in the low

and medium risk zones

- Compare the above behaviours and actions for elephants in the presence and absence of

identified disturbances (approach of vehicle, pedestrian or dog), for each risk zone

Furthermore, researchers investigated whether behaviour differed across males and females for each

risk zone, and in the presence of known disturbances. We also investigated the potential influence of

prior negative interactions with humans, using wounds and scars as a proxy.

Information generated from this study will assist in identifying how, or indeed if, Asian elephants are

altering certain behavioural patterns, when inhabiting areas closer to human communities, and how

they are responding to the associated perceived anthropogenic risks. Contributing to the knowledge

base of animal behaviour inside and outside of protected areas generally, and specifically in response

to anthropogenic disturbance and risk, facilitates comparisons across both sites and species, and

enables retro-active evaluation of the impact of human disturbance on elephants. The future well-

being of elephant populations depends at least partially on facilitating a more peaceful coexistence

between humans and elephants living in proximity to one another. The more information garnered

36

about general behaviour patterns in human dominated landscapes, and specific responses to

anthropogenic risk and disturbance, the better equipped conservationists, managers and other policy

makers are to develop management strategies with the welfare of both elephants and humans at the

forefront.

2.3 Methods

2.3.1 Study site

This study was conducted within the dry zone of Central Sri Lanka, at two neighbouring locations which

nonetheless varied in levels of anthropogenic disturbance. The Central Sri Lankan dry zone has an

average annual temperature of 28°C, with little fluctuation throughout the year, and is characterised

by two wet seasons: the ‘long rains’ from October to January, and the ‘short rains’ from March to May.

The driest time of year is between July and September (de Silva and de Silva 2007; Fernando et al.

2005; Wikramanayake et al. 2004).

The two specific locations chosen for this research were: a) inside the south-eastern boundary of

Wasgamuwa National Park (park entrance: 07°39.051' N, 080°55.633' E), and b) an area outside of

park boundaries and in between villages called ‘Weheragala Tank’ (Figure 2.1). Both of these locations

are characterised by tropical dry mixed evergreen forest habitat bordering grassy clearings, with

access to permanent, although fluctuating, water sources. Elephants are commonly observed in both

areas, and it was considered highly likely that many individual elephants within the overall population

would utilise both areas. This provided a unique opportunity to study the behaviour of elephants from

the one population in areas of similar habitat but with varying degrees of anthropogenic activity and

low and medium disturbance. Additionally, both sites were easily accessible by vehicle with clearings

providing relatively unobstructed and safe observation opportunities. Further information on each

location is detailed below.

37

2.3.1.1 Wasgamuwa National Park (low risk zone)

Wasgamuwa National Park spans 370 km² however much of the park is inaccessible either due to

terrain or distance from the one park entrance, and visitors typically use only the south-eastern

section which is near the Mahaweli River, and several man-made water tanks. The water sources and

grassy clearings throughout this section of the park provide a focal point for wildlife, including

elephants. The south-eastern boundary of Wasgamuwa National Park is partially surrounded by an

electric fence however, this is often poorly maintained, and elephants commonly use this area as an

access point between the park and neighbouring villages. Our data collection was concentrated in the

clearings observable from the many dirt roads, throughout an overall area of approximately 46 km² of

this south-eastern section of Wasgamuwa National Park (Figure 2.2).

Figure 2..1. Overview of study site: Wasgamuwa National Park (low-risk zone) and Weheragala Tank (medium-

risk zone). The green square highlights the south-eastern section of Wasgamuwa National Park encompassing

our observation area, while the orange circle represents Weheragala Tank and immediate surrounds.

38

Wasgamuwa National Park was defined as a ‘low-risk zone’ for elephants, as by law, human-elephant

interactions within park boundaries should be limited to rangers and visitors travelling in vehicles

along designated tracks and minimising behaviour that could disturb the elephants. Furthermore,

visitors are only permitted in the park between 7 am and 6 pm, and at all times elephants can seek

refuge in forest away from humans.

2.3.1.2 Weheragala Tank (medium risk zone)

Located approximately two kilometres from the nearest park boundary and six kilometres from the

park entrance, Weheragala Tank is situated between three villages, all which experience some degree

of human-elephant conflict in the form of crop-raiding (pers. comm). Weheragala Tank is

characterised by a large man-made water tank, surrounded by a grassy clearing and forest habitat

(Figure 2.3). Water levels in the tank fluctuate widely throughout the year, generally coinciding with

the rain seasons, although local government departments can release or take water at any time. At its

most full, the tank covers an area of approximately 0.254 km² with water reaching up to the forest

edge. At its driest, water spans an area of only approximately 0.045 km² and a grassy clearing of up to

250 m length separates the forest and tank edges. When the tank is full, the western and southern

sides are not accessible by vehicle.

Weheragala Tank was defined as a ‘medium risk zone’ for elephants, as humans and elephants both

use resources such as water, forest edges and paths or roads. Negative interactions do occur,

particularly in the early mornings or late afternoons. At this time, people are often traveling between

villages for school or work, while elephants are often emerging from or returning to forests, before or

after foraging in the clearings and drinking from the water tanks. However, elephants are not exposed

to the same high level of negative human-elephant interactions as when they directly enter villages to

forage on cultivated crops. Thus, the anthropogenic disturbance and risk presented at Weheragala

Tank falls somewhere in the middle of the scale between PA’s and high conflict zones.

39

Figure 2.3 Weheragala Tank ‘medium risk zone’, showing the water tank and surrounding forest.

Figure 2.2 South-eastern section of Wasgamuwa National Park,showing tracks driven during

observation sessions, water reservoirs and the Mahaweli River.

40

2.3.2 Study subjects

Subjects were 72 wild and free-ranging adult Asian elephants (Elephas maximus) (25 females, 47

males), from a population inhabiting areas inside and outside of the south-eastern boundaries of

Wasgamuwa National Park. No male elephants exhibiting outward signs of musth (their reproductive

state) were included in the study. Elephants were observed in group sizes ranging from a single

elephant to groups of up to 52 individuals. Following the protocol outlined in Chiyo et al. (2011), a

group was defined as a spatially cohesive and behaviourally coordinated aggregation of two or more

elephants. ‘Spatially cohesive’ meant individuals were aggregated within a radius of 100 m and

oriented in the same general direction while moving, and ‘behaviourally coordinated’ meant the

majority of the group performed similar actions and/or interacted with other group members.

Individual adult elephants were identified using a photographic identification file which I developed

using photographs captured between May 2014 and November 2016 (e.g. Figure 2.4). Prior to this

study, no photographic identification file or other record of individuals from the local elephant

population existed. Consequently, precise ages of individuals were unknown. Elephants were

recognised as adults using a combination of cues based on body size, degree of ear folding, and

depigmentation (de Silva et al. 2011; Avirazhagen and Sukumar 2008; Gunawardene et al. 2004).

Each adult elephant was also given an ‘Elephant-Human Interaction (EHI)’ rating, using the presence

and absence of prominent lumps and wounds on their body (including head, limbs and tail) as a proxy

for negative interactions with humans. As confirmed by elephant behaviour and veterinarian experts

in Sri Lanka, these lumps are typical of those inflicted by humans using guns (including homemade or

shrapnel guns) and other catapults, often to chase elephants away from their homes or property.

Elephants were given a low, medium or high EHI score depending on the number of lumps present. If

both sides of the body could not be observed (and the elephant had less than 25 lumps visible on one

side – in which case an automatic high score is awarded), no rating was given (Table 2.1; Figures 2.5,

2.6 and 2.7). Due to small sample sizes, elephants with low and medium EHI scores were grouped

together, forming a low/medium category which we termed ‘low’ for the purposes of analysis.

41

Figure 2.4 Example from Elephant ID File. Mohan: page 1. This page shows left and right profiles and a front-on

view of Mohan’s face. Distinguishable depigmentation patterns, notches and holes in both ears are very obvious.

42

Figure 2.4 Example from Elephant ID File. Mohan: page 2. This page shows his full body left and right profiles,

illustrating numerous scars and lumps scattered about his body, and also an easily distinguishable shape for the

tail hair. All these features, in addition to ear shape and depigmentation provide important clues when attempting

to recognise individual elephants in the field.

43

EHI

score

Description

Low Elephant has ≤ 5 lumps anywhere on its body, head, limbs or tail

Medium Elephant has 6 to 24 lumps present, generally scattered sporadically across body,

head, limbs and/or tail but can also be clustered together in one or two locations

High Elephant has 25 ≥ lumps, generally across all areas of body, head, limbs and tail

Table 2.1 Elephant-human interaction (EHI) ratings and definitions

Figure 2..5 Elephant with a high Elephant-Human Interaction rating. This is an adult male elephant named

Abeeshan. This picture shows the numerous large lumps, bumps and wounds present on just the left side of his

body. Overall, I counted 46 lumps on his body and rated him as having a high level of previous negative

interactions with humans (high EHI). © Kylie M. Butler.

44

Figure 2.6 Elephant with a medium Elephant-Human Interaction rating. This is an adult male elephant

names Jagulan. This picture shows several lumps scattered about the right side of his body, including

near his rump. Overall, I counted 18 lumps on Jagulan’s body (the right side being worse affected than

the left). Jagulan was rated as having a medium level of previous negative interaction with humans

(medium EHI). © Kylie M. Butler.

Figure 2.7 Elephant with a low elephant-human interaction rating. This is an adult male

elephant named Amila. The picture shows that the right side of his body and face is free of

lumps or wounds. No lumps were counted on either side of his body and I rated him as having

a low level of previous negative interaction with humans (low EHI) © Kylie M. Butler.

45

2.3.3 Data collection: elephant observations

This study was conducted over an 18-month period between May 2015 and June 2016. Elephant

observations and recording of video footage in Wasgamuwa National Park were performed between

May 2015 and June 2016 in Wasgamuwa National Park, and between May 2015 and October 2016 at

Weheragala Tank. Elephant observations were typically performed in Wasgamuwa National Park once

per week, and at Weheragala Tank three times per week, during three ten-week field seasons. The

difference in sampling effort was due to reduced accessibility of the park following or during heavy

rain, the availability of a reliable research vehicle, and the availability of one of our two regular park

trackers.

The precise methods of approaching elephants and determining observations spots differed between

the two locations. This was due to differences in size, with Wasgamuwa National Park being much

larger and requiring us to drive around in order to find elephants, and Weheragala Tank being much

smaller requiring only stationary observation spots. Furthermore, in Wasgamuwa National Park we

were required to remain inside a closed-roof vehicle at all times, and to drive only along designated

dirt roads (refer back to Figure 2.2). Whereas, at Weheragala Tank we were limited in where we could

drive, as few tracks existed in the forest. Furthermore, it was not safe to drive through the forest in

the late afternoon when elephants were active. Therefore, stationary observation spots in the clearing

were selected and we waited to see if elephants would appear, rather than actively searching for them

(refer back to Figure 2.3).

Observations in Wasgamuwa National Park took place between 3 pm and 6:30 pm (when the park

closed). The exact route and distance covered during each session depended upon weather, road

condition and how many elephants were observed. Often park staff had information about where

elephants were located, and as we aimed to collect as much footage of elephant behaviour as possible,

this also influenced the route. For personal safety, and out of respect for the elephants and to

minimise disturbance caused by our presence, we always tried to place ourselves more than 30 m to

the nearest elephant, and to park parallel to the group, switching the engine off as soon as possible.

However, on some occasions the elephants were not sighted until we were within 30 m due to thick

vegetation or bends in the road. Video recording began as soon as we were stationary and continued

for 20 minutes, or until the elephant moved out of sight (if less than 20 minutes). If it was unsafe or

would cause disturbance to the elephants to leave after 20 minutes (e.g. elephants were on the road,

46

elephants were close to the vehicle, or a group with juveniles were separated on both sides of the

road) video recording continued until it was deemed appropriate to leave (Figure 2.8).

Observations at Weheragala Tank took place between 3:30 pm and 6:30 pm, when it would start to

become too dark for filming. Whether we observed from set locations to the north of the tank or to

the south-west of the tank depended on the likelihood of elephants being present, as determined by

recent observations and information from residents of neighbouring villages. In the area to the north

of the tank, observations were performed from a 4.5 m high platform erected in a tree hut with a 360°

view (Figure 2.9). At the south-western location, observations were conducted from inside, or

standing next to, the research vehicle between the tank edge and the forest (at least 50 m from the

forest edge) (Figure 2.10). When the water tank was full, the areas to the south and west of the tank

were inaccessible, and all observations were done from the north. At Weheragala Tank, video

recording began when elephants were first sighted until elephants were out of sight, or it was too dark

to continue observing.

Figure 2.8 Elephant observations in Wasgamuwa National Park

Left: A group of elephants foraging on grass and interacting, as seen from research vehicle © Kylie M. Butler

Right: Researcher Kylie M. Butler photographing elephants in Wasgamuwa National Park © Supun Herath

47

Figure 2.9 Elephant observations to the north of

Weheragala Tank

Left: The Tree Hut platform used for elephant

observations. To the left of the tree hut is forest habitat

and to the right the water tank. © Lacy Wittman.

Right: Field assistant Supun Herath and researcher

Kylie Butler filming and photographing a group of

elephants from the tree hut. © Tim Hunt.

Figure 2.10 Elephant observations from south-west of Weheragala Tank

Researcher Kylie Butler film a group of elephants at a distance of approximately 100 m from atop

the research vehicle © Lacy Wittman.

48

All video footage was filmed by a trained research assistant or myself, using a Sony Handycam HDR-

CX240 HD, Canon Legria HFR706, and/or Panasonic Lumix DMC-T 280 Ultra Zoom digital camera. Every

attempt was made to video record all elephants in the group, with the aim of collecting data on as

many individual elephants as possible. However, this was dependent upon group cohesion, and

individual movement patterns during the observation session. For example, large groups often

scattered over a wide area to feed, some individuals would walk back into the forest and its vegetative

cover, or a group would surround the vehicle rendering it impossible to film all elephants.

Furthermore, elephants who ventured furthest from forest habitat and thus closest to our observation

spots, were the most easily visible to film. Due to restrictions on where we were allowed to drive in

Wasgamuwa National Park, and where it was safe to drive at Weheragala Tank, our ability to

manoeuvre into position to give all elephants equal opportunity to be filmed was limited. However,

at no stage was filming concentrated on any individual elephant, or group of elephants, because they

were seen to be doing something more active or ‘interesting’ than others.

While video footage was being recorded, another observer would record the data and time of the

observation, the number of elephant’s present, whether they were in an all-male, cow-calf or mixed

groups, and the predominant activity of the group. Field notes detailing general group behaviours and

interactions or reactions to external stimuli were also documented, and proved useful when cross-

checking with video footage for disturbance events.

2.3.4 Behaviour coding from video footage

Using field notes to help determine on which dates and when disturbances occurred during elephant

observations, video footage was viewed and categorised into five to ten-minute segments, and divided

into two contexts: a) baseline and b) disturbance. Baseline meant that no known disturbances

occurred during the selected video coding period, or five minutes before the beginning of the video

coding period, and that observed behaviour was therefore not influenced by any external negative

stimuli identifiable by the observer.

Disturbance was defined as when an event presumed to represent a perceived threat or disturbance

to elephants occurred. Such events included approach of another vehicle (safari jeeps in Wasgamuwa

National Park, or general vehicles traveling in between villages at Weheragala Tank including jeeps,

tuk tuks and motorbikes), people moving on foot within sight of elephants and approaching to less

49

than 50 m (Weheragala Tank only), and dogs approaching elephants to within 50 m (Weheragala Tank

only). All disturbance events were recorded opportunistically, and at no time did we encourage or

manipulate the onset of a disturbance.

Overall, video footage from 85 elephant sightings was recorded (30 from Wasgamuwa National Park,

55 from Weheragala Tank). From these 85 recordings, 34 videos had segments suitable for extraction

of behavioural data during the baseline context (14 Wasgamuwa National Park, 20 Weheragala Tank),

and 26 videos had segments suitable for extraction of behavioural data during the disturbance context

(8 Wasgamuwa National Park, 18 Weheragala Tank). Some videos were suitable for only one of

baseline or disturbance data, and some were suitable for both. When videos were suitable for both

contexts, baseline data was extracted from video segments occurring before the disturbance event,

or if this was not possible, a minimum seven-minute interval between the end of the disturbance

event, and the beginning of the video segment used in coding for baseline behaviour was applied.

For the baseline context, videos were deemed to be of suitable quality for analysis if a minimum of

one focal individual was visible for a minimum period of five minutes. For the disturbance context,

videos were deemed to be of suitable quality for analysis if a minimum of one focal individual was

visible when the disturbance occurred and then either remained visible in the clearing, or was

confirmed to be behind vegetation (e.g. was seen entering the forest, and video footage and/or field

notes confirmed he/she did not re-emerge) for a minimum five minute period following the onset of

the disturbance. For each context, videos were assigned a numerical code and order of video analysis

was selected randomly. Once a video was selected for analysis the group type, size and composition

were confirmed, and all focal individuals identified. Following this, behavioural data was extracted one

focal individual at a time, with order of focal individuals also selected randomly.

For each context, the time an elephant spent in locomotion, counted in seconds from the first step

taken to the last, was recorded and expressed as a percentage of overall time, to account for variations

in observation durations between focal individuals. Scan samples at 30 second intervals were also

recorded to estimate durations of feeding bouts and time elephants spent ‘standing still without

feeding’, the assumption being that if elephants were observed performing the same behaviour for

consecutive scan samples they were considered to have been doing so for the overall period. These

were also converted to percentages. Counts of other behaviours, specifically backing away, dusting,

50

Behaviour Operational definition

Back away Elephant backward locomotion of three paces or more, while facing and within

ten metres of a vehicle, person or other (non-conspecific) animal when

backwards locomotion begins

Dusting Elephant picks up dirt/dust with its trunk and throws dirt/dust anywhere over its

body, face, legs or behind ears

Feeding Elephant picks up vegetation with its trunk, then places vegetation into its mouth

and chews within 30-s of picking vegetation up. Between picking up vegetation

and placing it in its mouth, the elephant often spends time ‘preparing’ food. This

can involve swinging/flicking it from side-to-side and/or breaking it with foot, or

hitting it on the ground. Elephant often chews for a long time in between placing

more food in mouth.

Freeze Elephant stands with head, ears, trunk and body motionless for five seconds or

more. Posture is usually stiff.

Locomotion:

running

Fast-paced forward locomotion in a specific direction, five paces or more.

Elephants knees are noticeably raised from the ground as it moves. Tail is often

(but not always) help horizontal to the body or higher.

Locomotion:

walking

Elephant forward locomotion (without running) in a specific direction (left, right

or straight) five paces or more

Smell Elephant holds trunk in the air or extends trunk in front of face and above knee

height, with trunk tip pointed in a direction of interest for two seconds or more

Standing still Elephant stands still in one location without feeding (and without holding freeze

posture). Can include shuffling footsteps from side-to-side or back-and-forth.

Tail up Elephant holds its tail horizontal to body or higher in a stiff posture, for a

minimum of three seconds.

Trunk to

mouth (self)

Elephant touches trunk tip to its own mouth (when it is not feeding or drinking).

Vigilance Elephant stands with head raised and either held motionless or slowly scanning

from side-to-side, plus one or more of the following: ears held perpendicular to

body, tail base rigid, and/or base of trunk thrust forward. Posture is held for a

minimum of three seconds.

51

Table 2.2. Elephant ethogram. Developed, adapted and expanded from: Freeman et al. 2010; Horback et al.

2013, and researchers personal field observations of wild African and Asian elephants: Kylie Butler 2011 – 2016.

smelling, ‘tail up’, throwing objects, trunk touch to own mouth, and vigilance/freeze posture, were

also recorded (Table 2.2). These behavioural responses were selected as they are typical of those

performed when an elephant is exploring a perceived external stimulus or feeling stressed, and we

termed these ‘reactive behaviours’(Table 2.2). These were converted to a ‘rate per minute’ (RPM)

score, however counts of individual ‘reactive behaviours’ were very low. To account for low counts,

and increase comparability across risk zone, sex and context, all ‘reactive behaviours’ were grouped

together into a single count for each elephant, from which an RPM score was calculated.

2.3.5 Statistical analysis

The potential effects of context (baseline or disturbance) and the sex of elephants (female or male),

on: 1) % time spent feeding; 2) % time spent in locomotion; 3) % time spent standing still (without

feeding); and 4) reactive behaviours (RPM) were measured using linear mixed effects models (LMM).

For RPM scores, the duration of the video segment measured in seconds was weighted to account for

variations in observation time which could have a bearing on the number of reactive behaviours

performed. For each of the four behaviours of interest, the interaction between context and sex were

included as predictor variables, with the identity of individual elephants fitted as a random intercept

term to allow for the repeated measure (paired) comparison of individual elephants who were

observed in both contexts. Percentage time spent in locomotion, and percentage time spent standing

still (without feeding) were both log(10) transformed before analysis to improve normality of data

(Warton and Hui 2011). Significance of terms in the model (p-value = 0.05), and post-hoc tests of

differences in least-squares means between categories were assessed using the Satterthwaite

approximation of degrees of freedom obtained from the lmer package. The proportion deviance

explained by the fixed effects and combined random fixed effects respectively, were calculated using

the marginal and conditional r-squared (Nakagawa and Schielzeth 2013).

Although we were interested in whether elephants differed in their response to disturbance relative

to baseline according to sex as well as risk level, final sample sizes were too small to allow inclusion of

the three-way interaction (risk * context * sex) in statistical models. For this reason, we opted to

52

examine responses for each risk zone separately, including only the two -way (context * sex)

interaction.

Finally, the potential effects of prior negative interactions with humans (EHI levels) were tested for

male elephants only. This is because EHI levels were confidently determined for all males but not all

females. Furthermore, the effect of EHI on the behaviour of male elephants inside and outside of PA’s

is of particular interest, as male-biases in HEC events are common. As only two male elephants

received a low EHI score, these were grouped with ‘medium’ EHI elephants to form a combined

low/medium EHI category. To test for effects of EHI levels, we fitted linear mixed models for each

response variable with EHI level and context as explanatory variables. The potential effects of prior

negative interactions with humans (EHI levels) were also tested by fitting linear mixed models for each

response variable with EHI level and context as explanatory variables.

Statistical analyses were conducted with the R Studio Statistical Package 3.5.1 (R core team 2018).

LMM’s were constructed using the package ‘lme4’ (Bates et al. 2014), with model outputs generated

with the ‘effects’ package (Fox 2003) and visualised with ‘ggplot2’ (Wickham 2016).

2.4 Results

2.4.1 General behavioural activities

The time elephants spent engaged in the three main behaviours recorded during afternoon

observations barely deviated between Wasgamuwa National Park (low risk zone) and Weheragala

Tank (medium risk zone). At both locations, the predominant activity was feeding, and elephants (n =

72) spent approximately half of the observed time doing this. Time spent in locomotion and standing

still (without feeding) was approximately 20% for each, and for both locations. One-way between

group analyses of variance (ANOVA) on each activity separately showed no significant differences

between the mean time elephants spent in any of these activities between Wasgamuwa National Park

and Weheragala Tank (p = 0.05) (Table 2.3).

53

Wasgamuwa

National Park

Weheragala Tank

M (±SE) M (±SE) ANOVA

% time spent feeding 45.59 6.42 51.67 4.14 F(1, 96)=.592, p = .443

% time spent in

locomotion

18.76 4.24 19.88 2.67 F(1, 96)=.048, p = .826

% time spent standing still

(not feeding)

22.94 5.78 15.38 2.31 F(1, 96)=.781, p = .379

Table 2.3. % time spent in general behavioural activities: % time spent feeding, in locomotion and standing still

(without feeding) for elephants observed in Wasgamuwa National Park and at Weheragala Tank (mean times

and standard errors)

Despite this apparent alignment of the time spent in general behavioural activities at Wasgamuwa

National Park and Weheragala Tank, we were interested in investigating further to see whether

behaviours were uniform across genders, and also in different contexts within each location.

At Wasgamuwa National Park (low risk zone), we found no evidence that either sex or context,

influenced the amount of time elephants spent feeding, in locomotion, or standing still (without

feeding). However, analyses revealed a significant interaction between sex and context on time spent

standing still (Table 2.4, Figure 2.11), with females spending significantly more time than males

standing still during the disturbance context (males: M = 12.23, ± SE = 1.78, n = 4; females: M = 37.64,

± SE = 21.06, n = 4). It should be noted that sample sizes for each sex in the disturbance context at

Wasgamuwa National Park were small.

54

Response Predictor variable Coefficient SE t R²

(marginal)

R²

(conditional)

Locomotion

(log₁₀)

(Intercept) 0.93 0.14 5.56 0.06 0.75

Context – Disturbance 0.11 0.19 0.60

Sex – Female 0.29 0.24 1.21

Context – Disturbance: Sex

– Female

-0.21 0.29 -0.72

Feeding (Intercept) 54.13 9.65 5.61 0.05 0.53

Context – Disturbance -17.49 16.08 -1.09

Sex – Female -6.28 16.54 -0.38

Context – Disturbance: Sex

- Female

7.63 24.32 0.31

Standing still

(not feeding)

(log₁₀)

(Intercept) 0.82 0.22 3.8 0.15 0.98


Sex – Female -0.16 0.35 -0.45

Context - Disturbance: Sex

- Female

1.07 0.13 8.01

Table 2.4. LMM summary: effects of context * sex on general behaviour in Wasgamuwa National Park

Parameter coefficients, standard errors (SE) and t-values for the fixed effects in linear mixed models examining

the relationship between context * sex of elephants in the low risk zone, and R² values of the fixed terms

(marginal) and fixed and random terms (conditional). Coefficients for which the 95% confidence interval does not

include zero are shown in bold. Reference level of factors: context (baseline), sex (male).

55

At Weheragala Tank (medium risk zone), a significant effect of sex on both locomotion and feeding

was identified using the LMM (Table 2.5). For time spent in locomotion, the interaction between sex

and context was significant. Males spent significantly higher percentage of time than females in

locomotion during the baseline context (males: M = 18.53%, ± SE = 4.13, n = 27; females: M = 5.79%,

± SE = 1.20, n = 13) whereas females significantly increased their locomotion during the disturbance

context (M = 29.17%, ± SE = 7.44, n = 15), suggesting that females display more caution in response to

a disturbance than males (Figure 2.12). Both sex and context significantly influenced the time

elephants spent feeding, however no interaction between the two variables was apparent. Females

spent more time feeding than did males (females: M = 57.12%, ± SE = 6.39, n = 21; males: M = 49.60%,

Figure 2.11. % time spent standing still (without feeding) in Wasgamuwa National Park, and the influence

of sex and context. Female elephants spent significantly more time standing still (without feeding) during

the disturbance context, when in Wasgamuwa National Park (low risk zone)

56

± SE = 5.95, n = 33), and both sexes spent significantly less time feeding during the disturbance context

than during the baseline context (disturbance: M = 37.07%, ± SE = 6.28, n = 32; baseline: M = 63.35%,

± SE = 4.81, n = 40) (Figure 2.13). Finally, the percentage time spent standing still (without feeding) did

not differ across sexes or context, and the interaction between these terms was non-significant (Figure

2.14).

Response Predictor variable Coefficient SE t R2

(marginal)

R2

(conditional)

Locomotion

(log10) (Intercept) 1.03 0.10 10.80

0.15

0.44

Context - Disturbance 0.15 0.14 1.09

Sex - Female -0.38 0.17 -2.26

Context – Disturbance : Sex

- Female 0.47 0.22 2.16


Context - Disturbance -18.45 9.24 -2.00

Sex - Female 23.59 10.84 2.18

Context – Disturbance : Sex -

Female -24.54 14.62 -1.68

Standing Still

(not feeding)

(log10)

(Intercept) 1.12 0.12 9.70

0.09

0.09


Sex - Female -0.39 0.20 -1.94

Context – Disturbance : Sex -

Female 0.31 0.30 1.05

Table 2.5. Parameter coefficients, standard errors (SE) and t-values for the fixed effects in linear mixed models

examining the relationship between treatment * sex of elephants at WT (medium risk zone), and R2 values of the

fixed terms (marginal) and fixed and random terms (conditional). Coefficients for which the 95% confidence

interval does not include zero are shown in bold. Reference level of factors: Treatment (Baseline), Sex (Male).

57

Figure 2.12. % time spent in locomotion at Weheragala Tank (medium risk zone). Females increase time

spent in locomotion during a disturbance.

58

Figure 2.13. % time spent feeding at Weheragala Tank (medium risk zone). Females spent the most time

feeding but this was reduced during the disturbance context.

59

Figure 2.14. % time spent standing still (without feeding) (medium risk zone). Relationship between context and

sex for time spent standing still at Weheragala Tank (medium risk zone). Male elephants spend more time

standing still (without feeding) than do females.

60

2.4.2 Specific ‘reactive’ behaviours

At both Wasgamuwa National Park (low risk zone) and Weheragala Tank (medium risk zone), ‘rate per

minute’ scores of reactive behaviours were low (Wasgamuwa National Park: M = 0.33, ± SE = 0.11, n

= 21; Weheragala Tank: M = 0.24, ± SE = 0.05, n = 54), and there was no significant difference in

frequency of reactive behaviours between locations (F(1, 96) = .781, p = .379). However, at

Wasgamuwa National Park, the frequency with which reactive behaviours were observed was

significantly higher during the disturbance context (M = 0.77 , ± SE = 0.11, n = 8) compared to during

the baseline context (M = 0.15, ± SE = 0.06, n = 18). This effect was also present at Weheragala Tank,

where again more reactive behaviours were observed during the disturbance context (M = 0.44, ± SE

= 0.11, n = 32) relative to the baseline context (M = 0.07, ± SE = 0.02, n = 40) (Table 2.6).

Response Predictor variable Coefficient SE z R2

(marginal)

R2

(conditional)

Wasgamuwa National Park (low risk zone)

Reactive

behaviours (RPM)

(Intercept) -6.91 0.59 -11.72

0.15

0.33


Sex - Male 0.36 0.72 0.50

Context – Disturbance :

Sex - Female -1.43 0.77 -1.86

Weheragala Tank (medium risk zone)

Reactive

behaviours (RPM) (Intercept) -9.40 1.05 -8.93

0.27

0.49


Sex – Male 2.43 1.09 2.23

Context – Disturbance:

Sex – Male -2.63 1.09 2.41

Table 2.6. Relationship between sex, context and the interaction between sex and context for reactive

behaviours in Wasgamuwa National Park and Weheragala Tank. Parameter coefficients, standard errors (SE)

and z-values for the fixed effects in generalised linear models examining the relationship between context * sex

of elephants at Wasgamuwa National Park and Weheragala Tank, and R2 values of the fixed terms (marginal)

and fixed and random terms (conditional). Coefficients for which the 95% confidence interval does not include

zero are shown in bold. Reference level of factors: context (baseline), sex (female).

61

2.4.3 Influence of prior negative interactions with humans (EHI levels)

Unfortunately, for Wasgamuwa National Park (low risk zone) sample sizes were insufficient to perform

analysis comparing responses across context and EHI. Although the LMM suggests an interaction

between context and EHI level for time spent in locomotion, with only two elephants of high EHI in

each context, further sampling is required to draw conclusions. EHI levels alone did not significantly

effect the time spent in locomotion, feeding or standing still for male elephants observed in

Wasgamuwa National Park (Table 2.7), nor did it effect the frequency of reactive behaviours

performed (Table 2.8).


(marginal)

R2

(conditional)

Locomotion (log10)

(Intercept) 1.03 0.21 4.91

0.47

0.47


EHI - Low 0.08 0.25 0.31


EHI - Low 1.19 0.44 2.71



EHI - Low 7.56 24.50 0.31


EHI - Low -45.58 37.13 -1.23

Standing still (not

feeding) (log10) (Intercept) 0.50 0.41 1.21

0.04 0.98


EHI – Low 0.34 0.50 0.67


EHI – Low -0.09 0.22 -0.40


examining the relationship between treatment * EHI levels of male elephants in Wasgamuwa National Park and

R2 values of the fixed terms (marginal) and fixed and random terms (conditional). Coefficients for which the 95%

confidence interval does not include zero are shown in bold. Reference level of factors: Treatment (Baseline),

EHI (High)

62

Response Predictor variable Coefficient SE z R2

(marginal)

R2

(conditional)

Reactive

Behaviour (RPM) (Intercept) -8.57 1.50 -5.73

0.10

0.52

Treatment - Disturbance 2.04 0.82 2.49

EHI - Low 1.93 1.51 1.28

Treatment –Disturbance:

EHI - Low -1.87 0.97 -1.92

Table 2.8. Parameter coefficients, standard errors (SE) and z-values for the fixed effects in generalized linear

mixed models examining the relationship between treatment * EHI levels of elephants in the low risk zone, and



EHI (High).

Sample sizes were higher at Weheragala Tank (medium risk zone), and of the 33 individual male

elephants for whom behaviour was scored, 13 had a high EHI level and 19 fell within the combined

low/medium category. One elephant was excluded from this analysis as his EHI level could not be

determined. The percentage time spent in locomotion, feeding and standing still (without feeding)

differed significantly between elephants of the two EHI categories (Table 2.9). Male elephants with a

high EHI level spent more time in locomotion than those with a low EHI level (high EHI: M = 33.92, ±

SE = 6.78, n = 19; low EHI: M = 13.09, ± SE = 3.23, n = 19), but less time feeding (high EHI: M = 28.77, ±

SE = 7.65, n = 19; low EHI: M = 66.58, ± SE = 7.15, n = 19). Male elephants with a high EHI level also

spent more time standing still (without feeding) than those with a low EHI level (high EHI: M = 25.81,

± SE = 6.26, n = 13; low EHI: M = 11.20, ± SE = 2.13, n = 19). Furthermore, elephants with a high EHI

level were observed performing reactive behaviours more frequently (high EHI: M = 0.45, ± SE = 0.17,

n = 13; low EHI: M = 0.08, ± SE = 0.03, n = 19) (Table 2.10).

63


(marginal)

R2

(conditional)

Locomotion (log10)

(Intercept) 1.25 0.16 7.66

0.21

0.41


EHI - Low -0.34 0.20 -1.70

Context – Disturbance :

EHI - Low -0.33 0.29 -1.11



EHI - Low 27.59 13.08 2.11


EHI - Low 14.27 18.03 0.79

Standing still (not

feeding) (log10) (Intercept) 1.37 0.18 7.71

0.17

0.37


EHI – Low -0.37 0.22 -1.72


EHI – Low -0.04 0.32 -0.13


examining the relationship between treatment * EHI levels of male elephants at Weheragala Tank and R2 values

of the fixed terms (marginal) and fixed and random terms (conditional). Coefficients for which the 95% confidence

interval does not include zero are shown in bold. Reference level of factors: Treatment (Baseline), EHI (High).

64

Response Predictor variable Coefficient SE

R2

(marginal)

R2

(conditional)

Reactive

Behaviour (RPM) (Intercept) -5.89 0.36

-

16.26

0.24

0.39

Treatment - Disturbance 0.96 0.29 3.31

EHI - Low -1.86 0.54 -3.47

Treatment – Disturbance:

EHI - Low 0.33 0.62 0.53

Table 2.10. Parameter coefficients, standard errors (SE) and z-values for the fixed effects in generalized linear

mixed models examining the relationship between treatment * EHI levels of elephants at Weheragala Tank, and



EHI (High).

2.5 Discussion

2.5.1 General behaviour patterns in Wasgamuwa National Park and Weheragala Tank

This study revealed no evidence that elephants observed at Wasgamuwa National Park (low risk zone)

and Weheragala Tank (medium risk zone) differed in general behavioural patterns including time

spent in locomotion, feeding, or standing still (without feeding), or in the frequency of reactive

behaviours performed. Elephants spent most of their time feeding, which is not surprising considering

that to meet their nutritional requirements, Asian elephants consume up to 150 kg of vegetation every

day and often forage for up to 18 hours per day (Sukumar 1989; Vancuylenberg 1977, McKay 1973).

Although elephants are general feeders and consume a wide variety of plants, grass forms an

important component of their diet (Alahakoon et al. 2017; Sukumar 1989; Vancuylenberg 1977) and

both Weheragala Tank and the south-eastern section of Wasgamuwa National Park provide ample

opportunity for grazing. Elephants spent approximately 20% of their time in locomotion, generally

walking between clearings and the forest, or towards water tanks to drink. About the same proportion

of time was spent standing still (without feeding). Elephants very rarely performed ‘reactive’

behaviours.

65

The lack of difference in behaviour across the two risk zones, contrasts with numerous other studies

that have demonstrated changes in behaviour relative to the level of anthropogenic disturbance

present, for a wide array of species (Caro 2005; Lacy and Martins 2003; Frid and Dill 2002). Many of

these studies show a reduction in foraging activity at higher levels of anthropogenic disturbance

(Donaldio and Buskirk 2006; White et al. 1999) however, most studies do focus on more extreme

differences in risk across sites. However, in one study looking at the behaviour of large mammals inside

and outside of PA’s in Tanzania, elephants did not adjust their behaviour in the presence of a potential

anthropogenic disturbance, although most of the other species in the area did (Kiffner et al. 2014).

When an animal alters its usual behaviour as a response to negative external stimuli, costs to fitness

or health are often incurred (Blumstein et al. 2005; Lacy and Martins 2003; Frid and Dill 2002; Anthony

and Blumstein 2000; Lima 1998). This means that whether consciously or unconsciously, behavioural

choices are often governed by some level of cost-benefit analysis or trade-off (Bunnerfeld et al. 2006).

Given the abundance of available forage at Weheragla Tank, coupled with the shelter (plus additional

food sources) provided by the forest on the outskirts of the grassy clearings, elephants could in fact

be drawn to leave the protected boundaries of Wasgamuwa National Park because of the resources

available here. It could be that the ‘medium’ level of risk presented at Weheragala Tank is below the

threshold at which significant behavioural changes would be likely to occur, and that the benefits of

easily accessible food, water and shelter outweigh the slightly increased risk of interactions with

humans.

Thus, the lack of obvious behavioural differences between Wasgamuwa National Park and Weheragala

Tank could indicate that the level of risk represented by each site was not as divergent as presumed.

Whilst Weheragala Tank certainly appears to accurately represent an area of ‘medium’ anthropogenic

risk, there were indications that Wasgamuwa National Park did not fully represent the ‘low risk’ zone

it theoretically should be. Visitor numbers at Wasgamuwa National Park are considerably lower than

at other elephant inhabited parks in Sri Lanka (Senervirathna & Perera 2013), therefore elephants

within this population may be less habituated to tourists and consider their presence a risk or

disturbance, just as they would outside a PA. Furthermore, it was not uncommon to see vehicles

revving their engines or approaching elephants quickly, driving between mothers and offspring, and

talking loudly. These are all human behaviours which elephants could perceive as a disturbance or

threat (researcher pers. obs).

66

Identifying ‘medium’ risk zones near other National Parks within Sri Lanka and comparing behaviour

inside and outside of these PA’s would provide a useful comparison to our study in the Wasgamuwa

National Park region and provide further information as to how, or indeed if, Sri Lankan elephants

adjust their behaviour at an interim level of anthropogenic disturbance.

2.5.2 Effects of context and sex on elephant behaviour

Whether elephants were located in the low or medium risk zone did not appear to influence

behaviour, however the presence of an immediate anthropogenic risk or disturbance did. At

Weheragala Tank, we opportunistically witnessed disturbances including vehicles, pedestrians or dogs

approaching the elephants at a distance closer than 50 metres (Figure 2.15). Often people would be

revving engines of motorbikes or tuk tuks, playing loud music, or shouting loudly as they approached.

In response to this direct presence of a disturbance, elephants reduced the time they spent feeding

and increased the frequency of reactive behaviours. The most common reactive behaviours

performed were smelling (Figure 2.16) (26 times by 14 individual elephants) and assuming the

vigilance posture (16 times by 7 individual elephants), both of which are used to gather information

about the surrounding environment (Bates et al. 2007; King et al. 2007). Nine elephants were also

seen holding their tails in a stiff upright position), which is a posture commonly seen before elephants

flee. A sex-bias in behaviour was also observed with only female elephants increasing the time spent

in locomotion following a disturbance and being more likely to walk or run back towards the forest.

Inside Wasgamuwa National Park, the effect of disturbance was less marked, with no change in

feeding or locomotion for either sex, and only females altering time spent standing still, which

decreased during the disturbance context. As at Weheragala Tank, specific reactive behaviours

increased, and again smelling was the most common response. However, results from Wasgamuwa

National Park should be treated cautiously, as only eight elephants were able to be scored for

behaviour during a disturbance, and the only opportunistic disturbance available was the approach of

tourist vehicles (Figure 2.17).

67

Figure 2.16. Elephant smells towards the research vehicle in Wasgamuwa National Park (low risk zone) ©

Kylie M. Butler

Figure 2.15. Vehicles and elephants in Wasgamuwa National Park (low risk zone). A tourist vehicle approaches

an elephant to within 20 m without stopping © Kylie M. Butler

68

Given that research observations were conducted from a vehicle, the elephants from our Wasgamuwa

National Park (low-risk zone) sample were likely biased to individuals who were more comfortable in

the presence of vehicles than others. Of the 54 elephant groups encountered in Wasgamuwa National

Park during the study period, 17 groups (31.5%) either fled into the forest as we approached, charged

at the car, or both. However, most fleeing or charging occurred before we were able to set up

recording equipment thus we were unable to collect behavioural data on individuals within these

groups. Observations from hides at stationary locations throughout the park would enable less biased

sampling however we did not have permission to leave the vehicle or establish such look-outs. Future

research perhaps in collaboration with park staff is encouraged.

Akin to what was witnessed at Weheragala Tank, interruptions to foraging as a response to

disturbance are common among species exposed to the physical presence of humans through wildlife

tourism or hunting (e.g. bears (White et al. 1999; Olson et al. 2007), deer (de Boer et al. 2004), dolphins

(Lusseau 2003), whales (Lusseau et al. 2009; Williams et al. 2006)), and sex-biases in response also

commonly occur (Bunnerfeld et al. 2006; Saj et al. 1999). For most of these species, the potential long-

term effects of altered foraging patterns include loss of body condition and reduced health and fitness

Figure 2.17. Elephants fleeing from the clearing towards the forest at Weheragala Tank © Kylie M. Butler

69

(Blumstein et al. 2005; Lacy and Martins 2003; Frid and Dill 2002; White et al. 2002; Anthony and

Blumstein 2000; Lima 1998). For elephants specifically, female investment in offspring is high with an

18 to 20-month gestation period, followed by up to four years before a calf is completely weaned, and

the nutritional intake required to support this is high (Sukumar 1989). Male elephants also have high

energetic requirements relating to their reproductive states and success (Sukumar 1989).

Consequently, disruptions to foraging patterns could have costly effects on the health and fitness of

both female and male elephants. It is difficult, if not impossible, however, to conclusively determine

at what intensity changes to foraging would have to be for such consequences would occur. Identifying

long-term changes in body condition or reproductive fitness would take years, if not decades to

properly quantify, and accounting for confounding variables (e.g. food availability, seasonal changes

in the environment, and fluctuations in anthropogenic pressure) would be extremely difficult in the

wild. Still the facts remain that Asian elephants are listed as endangered on the IUCN Red List of

Threatened Species (Choudhury et al. 2008), their natural habitat is under ongoing threat (Fernando

et al. 2014; Haturusinghe and Weerakoon 2012; Campos-Arceiz et al. 2009; de Silva and de Silva 2007;

Fernando et al. 2005), and human-elephant conflict is threatening their long-term survival in the wild

(Fernando et al. 2011; Campos-Arceiz et al. 2009; Fernando et al. 2005). Given that significant changes

to feeding behaviour were evident, as a precautionary measure it would be best-practice to reduce

anthropogenic disturbance where possible. Educating people about the dangers of deliberately

disturbing elephants (e.g. by driving towards them, revving engines, or throwing firecrackers while

elephants are feeding peacefully) and perhaps imposing penalties on those found doing so, would be

a good first step.

Of additional concern, is whether elephants who have previous exposure to negative interactions with

humans, will increase displays of aggressive behaviour towards humans, as has been suggested by

some researchers (Fernando et al. 2011; Perera 2009). Most human-elephant interactions at

Weheragala Tank are incidental, occurring as people travel to or from villages (for work, school or

markets), and elephants are foraging in the clearings or walking from the forest to the water tank.

These interactions usually involve either (or both) parties receiving a fright, and generally humans

have to wait patiently for elephants to move on or drive slowly past when it is safe to do so. However,

intentional anthropogenic disturbances, where people rev engines, drive off-road towards elephants,

or throw missiles, are not uncommon, and elephants often respond with threat displays or by charging

(researcher pers. obs.). If elephants were to associate all people with these types of disturbances,

70

consequences for innocent villagers could be dire. Fortunately, we found no evidence that elephants

with high levels of prior negative interaction with humans, were responding to the observed

disturbances more reactively than others. Still, our proxy of using wounds and scars to determine

elephant-human interaction levels, may be more relevant in the specific context where these wounds

were inflicted, which is most likely inside villages while crop-raiding. In the interest of facilitating a

more peaceful co-existence between humans and elephants, in a world where increased interactions

are inevitable, we advocate showing basic respect to wild elephants by not intentionally disturbing

them, regardless of whether disturbing them increases aggression by elephants to humans, or not.

2.6 Conclusion

This study provides initial evidence that elephants inhabiting areas of ‘medium’ level anthropogenic

disturbance outside of Wasgamuwa National Park boundaries are reducing time spent feeding, as a

response to immediate anthropogenic threats in the environment. That elephants increased specific

‘reactive’ behaviours, including smelling and assuming the vigilance posture when vehicles

approached, both inside and outside of Wasgamuwa National Park further indicates that certain

human actions are viewed as a threat or disturbance by the elephants.

Additionally, this study highlights some of the difficulties involved in assessing behaviours of wild

animals in a wild context, according to a perceived level of anthropogenic risk. Although theoretically

the distinction in anthropogenic risk present inside and outside of PA boundaries should be clear, we

found that irresponsible behaviour of people in tourist vehicles inside Wasgamuwa National Park

often mirrored the behaviour of people outside the park, thus potentially blurring the distinction

between ‘low’ and ‘medium’ risk zones. Furthermore, when relying on opportunistic disturbances, it

can be difficult to attain sufficient sample sizes however it is the only ethical way to attempt this.

Undoubtedly, the question of influences of anthropogenic disturbance on wild elephant behaviour is

an interesting one. However, the time required to venture beyond initial evidence supporting

hypotheses that elephants do indeed alter their behaviour according to human-influenced activity,

and the difficulties in managing confounding variables, are significant challenges to research in the

wild. Nonetheless, further studies examining wild elephant behaviour across gradients of human

disturbance are fundamental in understanding the effect of anthropogenic disturbance and activity

71

on elephants living in close cohabitation with humans. That said, management plans for locations

such as the Wasgamuwa National Park/Weheragala Tank area, cannot afford to wait until longer-term

studies are complete. Management plans, to reduce anthropogenic disturbance where possible and

facilitate a safer environment for both humans and elephants where the two co-exist, should be

developed with the input of local authorities, researchers and conservationists, and local

communities. Management plans can then be informed and adapted, as research into this most

interesting yet complicated area progresses.

72

2.7 References

Alahakoon, AMDB, Pushpakumara, EMAB, Ellepola, G & Ranawana, KB 2017, ‘Food and feeding

patterns of Asian elephants in Uda Walawe National Park, Sri Lanka’, Gajah, vol. 46, pp. 4 – 13.



Anthony, LL & Blumstein, DT 2000, 'Integrating behaviour into wildlife conservation: the multiple ways

that behaviour can reduce Ne', Biological Conservation, vol. 95, pp. 303 – 315.

Bandara, R & Tisdell, C 2002, 'Asian elephants as agricultural pests: economics of control and

compensation in Sri Lanka', Natural Resources Journal, vol. 42, pp. 491 – 519.

Barnes, RFW, Dubiure, UF, Danquah, E, Boafo, Y, Nandjui, A, Hema, EM & Manford, M 2006, 'Crop-

raiding elephants and the moon', African Journal of Ecology, vol. 45, pp. 112-115.



doi:10.1016/j.cub.2007.09.060.

Benhaiem, S, Delon, M, Lourtet, B, Cargnelutti, B, Aulagnier, S, Hewison, AJM, Morellet, N &

Berheyden, H 2008, 'Hunting increases vigilance levels in roe deer and modified feeding site selection',

Animal Behaviour, vol. 76, pp. 611-618.

Blumstein, DT, Fernandez-Juricic, E, Zollner, PA & Garity, SC 2005, ‘Inter-specific variation in avian

responses to human disturbance’, Journal of Applied Ecology, vol. 42, no. 5, pp. 943 – 953.

Bunnefeld, N, Linnell, JDC, Odden, J, van Duijin, MAJ & Andersen, R 2006, 'Risk taking by Eurasian lynx

(Lynx lynx) in a human-dominated landscape: effects of sex and reproductive status', Journal of

Zoology, doi: 10.1111/j.1469-7998.2006.00107.x.

73

Burn, RW, Underwood, FM & Blanc, J 2011, 'Global trends and factors associated with the illegal killing

of elephants: a hierarchical Bayesian analysis of carcass encounter data', PLoS ONE, vol. 6, no. 9,

e24165. doi:10.1371/journal.pone.0024165.

Campos-Arceiz, A, Takatsuki, S, Ekanayaka SKK & Hasegawa, T 2009, ‘The human-elephant conflict in

Southeastern Sri Lanka: Type of damage, seasonal patterns, and sexual differences in the raiding

behaviour of elephants’, Gajah, vol. 31, pp. 5 – 14.

Caro, TM 2005, 'Behavioural indicators of exploitation', Ethology, Ecology and Evolution, vol. 17, no.

2, pp. 189-194.

Caro, TM 1999, 'Demography and behaviour of African mammals subject to exploitation', Biological

Conservation, vol. 91, no. 1, pp. 91 – 97.





9, no. 4, pp. 261 – 270.

Crosmary, WG, Makumbe, P, Cote, SD & Fritz, H 2012, 'Vulnerability to predation and water

constraints limit behavioural adjustments of ungulates in response to hunting risk', Animal Behaviour,

vol. 83, pp. 1367-1376.

Davidson, Z, Valiex, M, Van Kesteren, F, Loveridge, AJ, Hunt, JE, Murindagomo, F & Macdonald, DW

2013, ‘Seasonal diet and prey preference of the African lion in a waterhole-driven semi-arid savannah’,

PLoS ONE, vol. 8, no. 2, e55182. https://doi.org/10.1371/journal.pone.0055182.



doi: 10.1111/j.1755-263X.2011.00182.x

74

de Boer, HY, van Breikelen, L, Hootsmans, MJM & van Wieren, SE 2004, 'Flight distance in roe deer

Caperoulus capreolus and fallow deer Dama dama as related to hunting and other factors', Wildlife

Biology, vol. 10, no. 1, pp. 35 – 41.

de la Torre, S, Snowdon, CT & Bejarano, M 2000, ‘Effects of human activities on wild pygmy marmosets

in Ecuadorian Amazon’, Biological Conservation, vol. 94, pp. 153 – 163.

de Silva, M & de Silva, PK 2007, The Sri Lankan elephant: Its evolution, ecology and conservation, WHT

Publications, Colombo.

Donadio, E and Buskirk, SW 2006, 'Flight behaviour in guanacos and vicunas in areas with and without

poaching in western Argentina', Biological Conservation, vol. 127, pp. 139-145.



pp. 20 -25.

Fernando, P, de Silva, MKCR, Jayasinghe, LKA, Janaka, HK & Pastorini, J 2019, ‘First country-wide survey

of the endangered Asian elephant: toward better conservation and management in Sri Lanka’, Oryx,

pp. 1 – 10, https://doi.org/10.1017/s0030605318001254 .






Freeman, HD & Gosling, SD 2010, ‘Personality in non-human primates: a review and evaluation of past

research’, American Journal of Primatology, vol. 72, pp. 653 – 671.

Frid, A & Dill, L 2002, 'Human-caused disturbance: stimuli as a form of predation risk', Conservation

Ecology; vol. 6 (1), http://www.consecol.org/vol6/iss1.art11.

75




Graham, MD, Douglas-Hamilton, I, Adams, WM & Lee, PC 2009, 'The movement of African elephants

in a human-dominated land-use mosaic', Animal Conservation, vol. 12, pp. 445-455.

Griffin, SC, Valois, T, Taper, ML & Mills, LS 2007, 'Effects of tourists on behaviour and demography of

Olympic marmots', Conservation Biology, vol. 21, no. 4, pp. 1070-1081.



Gunn, J, Hawkins, D, Barnes, RFW, Mofulu, F, Grant, RA & Norton, GW 2013, ‘The influence on lunar

cycles on crop-raiding elephants: evidence for risk avoidance’, African Journal of Ecology, vol. 52, pp.

129 – 137.



Hollander, FA, van Overveld, T, Tokka, I & Matthysen, E 2008, ‘Personality and nest defence in the

great tit (Parsus major)’, Ethology, vol. 114, pp. 405 – 412.

Ihwagi, F, Thouless, C, Wang, T, Skidmore, AJ, Omondi, P & Douglas-Hamilton, I 2018, 'Night-day speed

ratio of elephants as indicator of poaching levels', Ecological Indicators, vol. 84, pp. 38 – 44.

Kahindi, O, Wittemyer, G, King, J, Ihwagi, F, Omondi, P & Douglas-Hamilton, I 2009, 'Employing

participatory surveys to monitor the illegal killing of elephants across diverse land uses in Laikipia-

Samburu, Kenya', African Journal of Ecology, doi: 10.1111/j.1365-2028.2009.01.200.x.

Kiffner, C, Kioko, J, Kissui, B, Painter, C, Serota, M, White, C & Yager, P 2014, 'Interspecific variation in

large mammal responses to human observers along a conservation gradient with variable hunting

pressure', Animal Conservation, doi: 10.1111/acv.12131.

76

King, LE, Douglas-Hamilton, I & Vollrath, F 2007, ‘African elephants run from the sound of disturbed

bees’, Current Biology, vol. 17, pp. R832 – R833.

Kumaraguru, A, Brinda, K & Asokan, S 2011, ‘Prey preference of large carnivores in Anamalai Tiger

Reserve India’, vol. 57, European Journal of Wildlife, pp. 627 – 637.

Lacy, KE & Martins, EP 2003, ‘The effect of anthropogenic habitat usage on the social behaviour of

vulnerable species, Cychura nubila’, Animal Conservation, vol. 6, pp. 3 – 9.

Lima, SL 1998, 'Nonlethal effects in the ecology of predator-prey interactions', Bioscience, vol. 48, no.

1, pp. 25 – 34.

Lima, SL and Dill, LM 1990, 'Behavioral decisions made under the risk of predation: a review and

prospectus', Canadian Journal of Zoology, vol. 68, pp. 619-640.

Loveridge, AJ, Hunt, HE, Murindagomo, F & Macdonald, DW 2006, ‘Influence of drought on predation

of elephant (Loxodonta africana) calves by lions (Panthera leo) in an African wooded savannah’,

Journal of Zoology, vol. 2770, no. 3, pp. 523 – 530.

Lusseau, D, Bain, DE, Williams, R & Smith, JC 2009, 'Vessel traffic disrupts the foraging behaviour of

southern resident killer whales Orcinus orca', Endangered Species Research, vol. 6, pp. 211-221.

Lusseau, D 2003, 'Male and female bottlenose dolphins Tursiops spp. have different strategies to avoid

interactions with tour boats in Doubtful Sound, New Zealand', Marine Ecology Progress Series, vol.

257, pp. 267 – 274.

Mattson, DJ, Blanchard, BM & Knight, RR 1992, ‘Yellowstone grizzly bear mortality, human

habituation, and whitebark pine seed crops’, The Journal of Wildlife Management, vol. 56, no. 3, pp.

432 – 442.

McKay, GM 1973, ‘Behaviour and ecology of the Asiatic Elephants in Southeastern Ceylon’,

Smithsonian Contributions to Zoology, vol. 125, Smithsonian Institution Scholarly Press, Washington,

D.C.

77

Okello, MM, Kiringe, JW, Njumbi, SJ and Isiche, J 2016, 'Prevalence of human-elephant conflicts in

Amboseli ecosystem, Kenya: current opinions of local communiy', International Journal of Biodiversity

and Conservation, vol. 8, no. 3, pp. 60 – 71.

Olson, TL, Gilbert, BK & Squibb, RC 1997, ‘The effects of increasing human activity on brown bear use

of an Alaskan river’, Biological Conservation, vol. 82, pp. 95 – 99.





pp. 4 – 11.

Power, RJ & Compion, RXS 2009, ‘Lion predation on elephants in the Savuti, Chobe National Park,

Botswana’, African Zoology, vol. 44, no. 1, pp. 36-44.

R Core Team 2018, R: A language and environment for statistical computing, R Foundation for

Statistical Computing, Austria.

Rangarajan, M 1996, ‘The politics of ecology: the debate on wildlife and people in India 1970 – 1995’,

Economics of Politics Weekly, vol. 31, pp. 2391 – 2409.

Saj, T, Sicotte, P & Paterson, JD 1999, ‘Influence of human food consumption on the time budget of

vervets’, International Journal of Primatology, vol. 20, pp. 977 – 994.



Senevirathna, HMMC & Perera, PKP 2013, ‘Wildlife viewing preferences of visitors to Sri Lanka’s

National Parks: implications for visitor management and sustainable tourism planning’, Journal of

Tropical Forestry and Environment, vol. 3, no. 2, pp. 1 – 10.

78

Setsaas, TH, Holmern, T, Mwakalebe, G, Stokke, S & Roskaft, E 2007, 'How does human exploitation

affect impala populations in protected and partially protected areas? – A case study from the Serengeti

Ecosystem, Tanzania', Biological Conservation, vol. 136, pp. 563-570.



Sitati, NW, Walpole, MJ, Smith, RJ & Leader-Williams, N 2003, ‘Predicting spatial aspects of human-

elephant conflict’, Journal of Applied Ecology, vol. 40, pp. 667 – 677.

Soltis, J, King, LE, Douglas-Hamilton, I, Vollrath, F & Savage, A 2014, ‘African elephant alarm calls

distinguish between threats from humans and bees’, PLoS ONE, vol. 9, no. 2, e89403.

doi:10.1371/journal.pone.0089403.

Steidl, RJ and Anthony, RG 2000, ‘Experimental effects of human activity on breeding bald eagles’,

Ecological Applications, vol. 10, no. 1, pp. 258 – 268.

Sukumar, R 1989, The Asian elephant: ecology and management, Cambridge University Press, USA.

Sukumar, R & Gadgil, M 1988, ‘Male-female differences in foraging on crops by Asian elephants’,

Animal Behaviour, vol. 36, no. 4, pp. 1233 – 1235.

Thuppil, V and Goss, RG 2013, 'Wild Asian elephants distinguish aggressive tiger and leopard growls

according to perceived danger' Biology Letters, vol. 9, 20130518,


Vancuylenberg, BWB 1977 ‘Feeding behaviour of the Asiatic elephant in South-East Sri Lanka in

relation to conservation’, Biological Conservation, vol. 12, no. 1, pp. 33 – 54.



243 – 251.

79






Wilkie, RD & Douglas-Hamilton, I 2018, ‘High-resolution tracking technology reveals distinct patterns

in nocturnal crop-raiding behaviour of an African elephant (Loxodonta africana) in Amboseli, Kenya’,


Williams, RM 1994, Behavioural responses of killer whales to whale-watching: opportunistic

observations and experimental approaches, Masters Thesis, Department of Zoology, University of

British Colombia.

White, D, Kendall, KC and Picton, HD 1999, 'Potential energetic effects of mountain climbers on

foraging grizzly bears', Wildlife Society Bulletin, vol. 27, no. 1, pp. 146-151.

80

Chapter 3: Crop-raiding in a high human-elephant conflict location in Sri Lanka: an assessment of the characteristics and patterns of crop-raiding in Dewagiriya Village

81

3.1 Abstract

Crop-raiding is a major form of human-elephant conflict, threatening the long-term viability of

elephant populations and the livelihoods of communities living in proximity to both Asian and African

elephants. In Sri Lanka, crop raiding has been identified as a predominant threat to elephant

conservation. We conducted an intensive three-year study into the crop-raiding patterns and

predictors of elephants entering a small rural village called Dewagiriya on the outskirts of Wasgamuwa

National Park. We found crop-raiding to be prevalent, occurring year-round with no clear seasonal

patterns. Male elephants either alone or in small groups were the most likely to crop-raid, and almost

exclusively at nights. Homes and crop fields closest to the water tanks and forest edges were most

vulnerable to elephant damage. We suggest that to best protect the community from elephant

damage, crop-guarding efforts be concentrated in these ‘higher’ risk areas.

3.2 Introduction

The facilitation of peaceful co-existence between humans and wildlife is an ongoing challenge for

conservation scientists, wildlife managers and communities worldwide (Goswami and Vasudev 2017).

In many locations, agriculture and development has destroyed or fragmented traditional wildlife

habitat. A reduction of suitable habitat in conjunction with human population growth and more

people being encouraged to settle near wildlife areas, has resulted in intensifying levels of human-

wildlife conflict (HWC) over resources including space, water and food (Puyravaud et al. 2019;

Manfredo 2015; Harich et al. 2013; Redpath 2013).

Crop-raiding, where wildlife forages on, or otherwise damages, agricultural crops, is a major source of

HWC (Saraswat et al. 2015; Chakraborty and Modal 2013; Ekanayake et al. 2011; Thapa 2010), and

sadly it is often the most vulnerable human and wildlife communities that are most negatively

impacted by this. In many areas where crop-raiding is a predominant issue, local communities are of

low socio-economic status relative to the rest of the country, and people struggle with the daily

realities of living alongside certain wildlife species (Barua et al. 2013; Ogra 2008; Rao et al. 2002). This

is particularly so for families who are reliant upon income generated from farming for basic living

expenses (e.g. health and education) and where the crops they grow are their main source of food

(Barua et al. 2013; Ogra 2008; O'Connell-Rodwell et al. 2000).

82

On the other side, although animals of varied species, shapes and sizes are known to cause havoc by

crop-raiding, those which are large-bodied and capable of inflicting physical injuries on humans, are

typically at the forefront of HWC concerns. Examples include both African and Asian elephants,

Loxodonta africana and Elephas maximus, (Hoffmeier-Karimi and Schulte 2014; Santiapillai et al.

2006); Indian rhinoceros, Rhinoceros unicornis, (Thapa 2010) and several primate species (e.g.

chimpanzees, Pan troglodytes: McLennan and Hill 2012; Mountain gorillas, Gorilla beringei: Hockings

and Hunle 2009; Sumatran orang-utans, Pongo abelii: Campbell-Smith et al. 2010). All these

aforementioned species are globally recognised as requiring protection to ensure their long-term

survival in the wild (IUCN Red List of Threatened Species), and unfortunately HWC (specifically crop-

raiding) presents yet another challenge to their future existence.

Crop-raiding by both African and Asian elephants has been a hot topic of conservation and

humanitarian concern for several decades, yet shows little sign of abating (Davies et al. 2011; Perera

2009; Sitati and Walpole 2006). Globally hundreds of human and elephant injuries and deaths are

reported each year (Thirgood et al. 2005), and farmers continue to experience severely reduced crop

yields and damage to property as a result of elephant visits (Okello et al. 2016; Amwata and Mganga

2014; Gubbi 2012; Ekanayaka et al. 2011; Campos-Arceiz et al. 2009). As farmers become more

desperate to defend their livelihoods, they often resort to more violent and harmful means of

deterring elephants, which can perpetuate the problem they are trying to alleviate. In Asia, where an

estimated 95% of traditional elephant habitat has been lost primarily to agriculture (Sukumar 2003;

2006; Hedges et al. 2005), human-elephant conflict (HEC) influenced by crop-raiding is recognised as

the biggest threat to the survival of wild elephant populations (Perera 2009; Sukumar 2006; Fernando

et al. 2005; Hedges et al. 2005; Leimgruber et al. 2003).

This complicated scenario is perhaps no better exemplified than by the crop-raiding issues identified

in Sri Lanka. With an estimated elephant population of over 5,800 wild elephants, Sri Lanka is home

to more than 10% of the global wild Asian elephant population and has the highest density of Asian

elephants (Fernando et al. 2011). Although Sri Lanka has more than 8,500 km² (13%) of land

designated as Wildlife Protected Area's (PA's) (DWC 2017), attempts to restrict wild elephants within

PA's have largely failed (Fernando et al. 2011; 2005) and it is not uncommon for elephants to be found

in higher densities outside of PA's with the majority of elephants spending a proportion of their time

in close proximity to rural communities (Fernando et al. 2011; Campos-Arceiz et al. 2009). Elephant

83

inhabited areas, including those on the periphery of villages, are being decimated by ongoing

deforestation for development, including dams and agricultural expansion (Fernando et al. 2014;

Haturusinghe and Weerakoon 2012; Campos-Arceiz et al. 2009; de Silva and de Silva 2007; Fernando

et al. 2005). This means that humans and elephants are being forced to meet their often overlapping

resource needs in increasingly close proximity. Consequently, crop-raiding appears to be worsening

and is having an overwhelming negative influence on the lives of many farmers, with the resulting

conflict the principle threat to the survival of Sri Lanka's elephants (Fernando et al. 2011; Campos-

Arceiz et al. 2009; Fernando et al. 2005).

Understanding how, when and why some elephants crop-raid is essential to predicting the

vulnerability of farms to crop depredation, and to identifying and implementing appropriate

mitigation strategies (Webber et al. 2011; Sitati et al. 2005). Predicting factors vary, however across

the vast global spatial scale in which elephant crop-raiding occurs, some general patterns have

emerged, and these can provide a useful comparative basis when investigating specific crop-raiding

conflict situations.

In many locations throughout Asia, elephants raid crops year-round with fruits (e.g. banana, coconut,

watermelon), vegetables (e.g. eggplant, pumpkin, sweet potato), legumes and grains (e.g. cowpeas,

maize, mung beans) and sugar cane commonly consumed (Chen et al. 2016; Ekanayaka et al. 2011;

Fernando et al. 2011; Webber et al. 2011; Barnes et al. 2005). Annually, seasonal variations in

frequency and intensity of crop-raiding are common, usually peaking towards the end of rain seasons

when major crops such as rice or maize are maturing (Chen et al. 2016; Das et al. 2014; Ekanayaka et

al. 2011; Webber et al. 2011). Within the 24-hour daily cycle, crop-raiding occurs almost exclusively at

night (Das et al. 2014; von Gerhardt et al. 2014; Graham et al. 2010; Sitati et al. 2003) yet typical

elephant feeding patterns show that the majority of foraging occurs during hours of daylight (Sukumar

2006). This tendency to crop-raid nocturnally indicates that elephants possess an awareness of the

risks associated with crop-raiding and adjust their behaviour accordingly (Gunn et al. 2013; Graham

et al. 2009, Barnes et al. 2006). Further supporting this risk-avoidance theory, at multiple locales in

Africa, the light afforded by lunar phase was found to influence crop-raiding activity. During new lunar

phases when nights are at their darkest, the occurrence of crop-raiding increased, and during full

moon phases when night-time visibility is greatest, crop-raiding decreased (Gunn et al. 2013; Barnes

et al. 2006). This has yet to be investigated in Asia.

84

Although both male and female elephants are often found outside of PA's and in proximity to villages,

sex-biases in crop-raiding are evident, with male elephants raiding far more frequently than females,

and typically doing so alone or in small bachelor groups (Thaufeek et al. 2014; Das et al. 2012; Sukumar

and Gadgil 1988). This suggests that in most locations, male elephants are more willing than female

elephants to exploit crops as part of a high-risk, high-gain foraging strategy (Sukumar and Gadgil 1988).

In Sri Lanka, where the importance of mitigating HEC is well-recognised, only seven peer-reviewed

papers specifically investigating predictors and patterns of crop-raiding behaviour at various conflict

locations throughout the country, have been published since 2002 (see Thaufeek et al. 2014;

Haturusinghe and Weerakoon 2012; Ekanayaka et al. 2011; Campos-Arceiz et al. 2009; Fernando et al.

2005; Wikramanayake et al. 2004; Bandara and Tisdell 2002). To the best of my knowledge, these

studies provide the only peer-reviewed data from which to compare crop-raiding characteristics in Sri

Lanka with other HEC zones globally, and for highlighting commonalities and differences in crop-

raiding within Sri Lanka. However, given the likelihood of a continuing increase in HEC throughout the

country (Fernando et al. 2011; Campos-Arceiz et al. 2009; Fernando et al. 2005), further research from

a wider array of conflict sites would contribute to the establishment of effective mitigation methods

by enhancing our understanding of the influencing factors and patterns of crop-raiding behaviour.

Almost all food crops grown in Sri Lanka are vulnerable to elephant damage (Thaufeek et al. 2014;

Haturusinghe and Weerakoon 2012), and at all sites assessed except one, various crops were subject

to some degree of elephant depredation throughout the entire year (Thaufeek et al. 2014;

Haturusinghe and Weerakoon 2012; Ekanayaka et al. 2011; Campos-Arceiz et al. 2009;

Wikramanayake et al. 2004; Bandara and Tisdell 2002). Rice, which is the country's primary agricultural

crop, was the most heavily targeted, and seasonal peaks in overall raiding coincided with the

approaching maturation of rice crops, typically beginning towards the end of the main rain season in

January (Thaufeek et al. 2014; Haturusinghe and Weerakoon 2012; Ekanayaka et al. 2011; Campos-

Arceiz et al. 2009; Fernando et al. 2005; Wikramanayake et al. 2004; Bandara and Tisdell 2002). At two

locations, a second crop-raiding peak was observed in August. These locations were in south-eastern

Sri Lanka and during this peak, permanent crops such as bananas, were the most frequently consumed

(Ekanayaka et al. 2011; Campos-Arceiz et al. 2009). High levels of attacks on houses, presumably so

elephants could access stored food inside, were evident in north-western, south-eastern and central

85

Sri Lankan conflict zones (Thaufeek et al. 2014; Haturusinghe and Weerakoon 2012; Campos-Arceiz et

al. 2009).

As with crop-raiding activity in other Asian and African locations, elephants in Sri Lanka used this

foraging strategy nocturnally and were rarely, if ever, seen in or approaching fields during daylight

hours (Campos-Arceiz et al. 2009; Wikramanayake et al. 2004; Bandara and Tisdell 2002). Male

elephants, either alone or in small all-male groups, were responsible for the majority of crop-raids,

and groups containing adult females raided far less frequently (Thaufeek et al. 2014; Haturusinghe

and Weerakoon 2012; Ekanayaka et al. 2011; Bandara and Tisdell 2002). Still, sex-biases in raiding

behaviour were not uniform across locations. At one site in north-central Sri Lanka, both male

elephants and female-led groups raided frequently, whereas at another site in southern Sri Lanka no

females at all were identified as raiders (Wikramanayake et al. 2004).

The Sri Lankan studies described above demonstrate broad patterns in targeted crop types, seasonal

and diurnal crop-raiding peaks, general sex-biases and group types of raiding elephants, yet also show

that even within a relatively small country, site-specific differences in crop-raiding do occur.

Furthermore, crop-raiding patterns or characteristics can differ within even an individual conflict

location. Very few studies in Africa or Asia provide an in-depth account of within-site variations (but

see: Chen et al. 2016; Barnes et al. 2005; Sitati et al. 2005), yet factors such as the proximity of a farm

to forest habitat or water sources, number of neighbouring farms or homes, plus socio-demographic

and economic factors, can all influence a farms vulnerability to crop-raiding (Chen et al. 2016; Sitati et

al. 2005; Hill 2004).

To properly identify the driving factors and likely consequences of crop-raiding, one cannot simply

look at patterns or predictors from other conflict zones, even within the same country or district. Such

information can provide a useful platform to build upon, however if the goal is to help communities

successfully alleviate HEC, the importance of identifying site-specific and within site characteristics

cannot be understated (Barnes 2009; Hill 2004).

In this study, we provide an in-depth investigation into the patterns and predictors of crop-raiding in

a single conflict location within a high HEC region of Sri Lanka. Specifically, we aimed to:

86

- quantify the frequency of elephant sightings in fields and home gardens and to describe the

patterns of damage incurred

- identify temporal patterns in crop-raiding at seasonal, lunar and diurnal scales

- determine sex-biases in raiding behaviour, as well as group type and size

Furthermore, we investigated whether significant within-site variations were present and the

potential spatial (e.g. location of field/home in relation to main road and forest periphery/water

tanks), socio-demographic and economic (e.g. household size, income) factors influencing these.

Information generated from this study will help to predict vulnerability of crops and homes to

elephants throughout the overall site and at individual farms. This can be used to develop mitigation

strategies tailored to specific temporal and spatial variations, hopefully encouraging the concentration

of oft-limited resources where they are most required, and will be most effective, for individual

farmers and the overall community. Additionally, we consider our findings in light of patterns and

predictors identified in existing crop-raiding studies in Sri Lanka and other affected areas. The more

information procured for a wider array of individual conflict locations, the more meaningful cross-site

comparisons will become in identifying and developing site-appropriate and targeted mitigation

strategies, based on commonalities and differences.

3.3 Methods

3.3.1 Study site

This study was conducted in the rural community of Dewagiriya Village, located in the dry zone of

Central Sri Lanka (Figure 3.1). Overall, the dry zone encompasses approximately 60%, or almost 40,000

km², of Sri Lanka (Fernando et al. 2011), spanning the north to central provinces then stretching south

down the east coast. The dry zone typically experiences annual rainfall between 1650 – 2100 mm, and

is characterised by two wet seasons: the 'long rains' from October to January, and the 'short rains'

from March to May (Fernando et al. 2005; Wikramanayake et al. 2004). The driest time of the year is

between July and September. The average annual temperature is 28°C, with little variation throughout

87

Figure 3.1: Location of study site. Map showing location of our study site, Dewagiriya Village, in the Sri Lankan dry

zone, and relative to Wasgamuwa National Park (WNP) boundaries and the Mahaweli River.

88

the year (de Silva and de Silva 2007). Wild elephants inhabit almost this entire zone, and are commonly

found both in and outside of designated PA's (Fernando et al. 2011).

Since the mid 1900's, the environment of the north-central and central provinces has been

dramatically altered by large scale development (Wikramanayake et al. 2004). Major hydro-projects,

particularly along the Mahaweli River, have seen the construction of many dams and reservoirs, and

facilitated the widespread expansion of irrigated agriculture. Large tracts of forest, which were

historically prime elephant habitat, have been cleared and converted into agricultural fields (de Silva

and de Silva 2007; Fernando et al. 2005). As the forests diminish, elephants are forced into smaller

areas of suitable habitat. At the same time, new human communities are being established and farms

are expanding closer to both the remaining forest and PA boundaries (de Silva and de Silva 2007).

Many of these projects continue today and increasing overlap of human and elephant communities is

inevitable.

The specific study location of Dewagiriya Village (07°33.928' N, 080°56.866' E) was established in 1958

(A. Dingira, pers. comm.), and is situated approximately ten kilometres below the southern boundary

of Wasgamuwa National Park, and two kilometres to the west of the Mahaweli River (Figure 3.1).

Although exact population numbers are unknown, at least 400 elephants inhabit Wasgamuwa

National Park and/or utilise the numerous forest patches, river banks and small reservoirs, in between

the southern park boundaries and Dewagiriya Village (researcher pers. obs. 2014 - 2016).

Covering an area of approximately 2 km², Dewagiriya village is home to around 40 permanent

households, with houses and fields scattered on either side of a dirt road running east to west through

the centre of the village. Four semi-permanent man-made reservoirs (known as ‘tanks’) are located in

the south area and are bordered by forest (Figure 3.2). These forests provide an excellent, albeit small,

habitat for elephants who often seek refuge there during the day, moving into the village to access

water sources and crops at night (BKG. Jayasena, pers. comm.).

The community is of a low socio-economic status and individual small-scale agriculture is the main

livelihood. Most families have a small garden next to their house of approximately ¼ acre where fruits

trees (e.g. banana, coconut, papaya), and small vegetable plots (e.g. long beans, eggplant, okra) are

grown. Each family also has an adjoining field, typically between one to three acres although

89

neighbouring fields often border each other, making the overall continuous field area much larger.

Rice is the predominant crop type cultivated, and is used for both subsistence and income generation.

Almost every family plants rain-fed rice fields during the 'long rains' and those with access to irrigated

water supplies will often plant a second rice crop during the 'short rains'.

At an initial community meeting I attended in June 2014, the 20 participants unanimously agreed that

HEC is a pressing issue for the entire community. Given the villages proximity to elephant habitat and

water sources, and the communities dependence on rice cultivation – a frequently raided crop in many

high HEC zones worldwide – Dewagiriya Village was deemed an ideal location for a long-term,

intensive investigation into the characteristics and patterns of crop-raiding.

3.3.2 Data Collection

This study was conducted over a three-year period from January 01st 2015 to December 31st 2017,

using resident reports of 'elephant events' collected during frequent home visits. I defined an

'elephant event' as any sighting of an elephant within the village, as well as any successful crop raids

that occurred.

There were a total of 40 households residing in the village at the conclusion of the study, however

four of these families had moved in part way through, one family lived in the village only during crop

growing seasons, and a further eight families were not home or contactable often enough to provide

consistent data. A household was deemed as not providing consistent data if I was not able to

interview them for a consecutive period of three months or more at any stage of the data collection

period. These thirteen households were excluded from analysis, leaving 27 participating households

scattered throughout the village (Figure 3.2).

Initially I distributed data sheets (translated into Sinhalese) to each household and trained an adult

family member how to answer each question. Each household was then visited every two to four

weeks to collect data sheets, discuss any elephant visits reported, and distribute new data sheets.

However, it was soon found that the majority of households failed to fill out their data sheets and

instead preferred to relay information directly during visits. To accommodate for this preference by

households, myself or my research assistants would scribe the participant's information directly onto

90

data sheets during the visits, with bilingual local field assistant Mr. Supun Herath translating from

Sinhalese into English.

For each elephant event the following information was recorded: household name and code; date of

sighting/crop raid; time of event (to the closest estimate); number of elephants; sex of elephants (or

unknown); location (field/home); whether damage was incurred and if so, a brief description; methods

used to deter elephants (if any); and any additional comments. Where substantial damage occurred

and the household was visited before damage was cleared or rectified, we inspected the damage with

the respondent, to verify reports and to gain insight into the nature of the HEC events experienced by

the community. However, I did not attempt to record precise measurements of damage or to estimate

economic losses due to the difficulties associated with assessing growth stage of crops, determining

whether damage was recoverable, and predicting projected crop yields in the absence of elephants

(e.g. Santiapillai et al. 2010; Thouless 1994).

I was also interested in potential influences of moon phase and rain on elephant events and for all

dates where an elephant event was reported, I recorded the moon phase and rainfall (mm). Moon

Figure 3.2: Map of the Dewagiriya Village study area.This map shows the location of houses, 'main' roads, and

water tanks. Houses designated by red diamonds indicate participants in our crop-raiding surveys.

91

phase data was collected from the freely available ‘Time and Date’ website (www.timeanddate.com)

and was categorised as full, waxing, new or waning following the protocol described in Gunn et al.

(2013) (Table 3.1). Presence or absence of rain (hereafter ‘raid day rain’) plus the cumulative total of

the previous 28 days rain (hereafter ‘previous month rain’) were recorded using rainfall

measurements (mm) purchased from the Department of Meteorology, Sri Lanka, however were

unavailable for December 2014 and July 2016. This unfortunately meant that I could not calculate

cumulative rainfall between the 1st and 28th of January 2015, and between the 1st July and 28th August

2016, and these dates were excluded from analysis.

Moon phase category Definition

Full Moon Night of the full moon, plus three nights before and three nights after

New moon Night of the new moon, plus three nights before and three nights after

Waning moon Nights between the full moon and the new moon (as defined above)

Waxing moon Nights between the new moon and the full moon (as defined above)

Table 3.1. Moon phase definitions

To facilitate a comparison of dates when elephant events did and did not occur, I generated an equal

number of randomly-chosen dates when elephants were not present in the village, with the

RANDBETWEEN command in Microsoft Excel using Julian dates for the duration of the study period,

and recorded moon phase and rainfall data for these also.

On any given date, it was possible for elephants to travel to multiple fields or household areas. As I

could not individually identify most elephants entering the village, I could not determine with certainty

if elephant events reported by multiple households on the same date were caused by the same

elephant, or group of elephants, or if separate groups of elephants were responsible. Thus, in order

to identify the frequency of elephant events throughout Dewagiriya village, the protocol described in

Gunn et al. (2013) was followed, where a 'raid day' period (24-hour period starting at 6 am) was used

to combine all records of elephant sightings and crop-raids within the village into a single count. The

majority of the time, reports from a given 'raid day' period consisted of the same or similar counts of

elephants, often concentrated in a localised pocket of the village. Based on this, the assumption that

multiple elephant events reported on any given 'raid day' period constitute only one event is

considered reasonable, while also avoiding the potential problem of over-estimating frequency of

elephant presence.

http://www.timeanddate.com/

92

I did however count the frequency with which elephants visited each household's property, to identify

potential HEC 'hotspots' throughout the village. To help determine potential influencing factors of

crop-raiding vulnerability within Dewagiriya Village, the location of each household in relation to the

main road (north or south of, and adjacent to main road or not) was recorded. This provided a measure

of proximity to forest/water tank periphery, as the south of the village and specifically those houses

not located adjacent to the main road, were closer to the forest and water tanks. I used this measure

rather than distance in metres due to the changeable nature of water tank boundaries influenced by

rainfall and irrigation. Additionally, the number of adults per household and estimated annual

household income was documented during semi-structured one-on-one interviews which were used

to assess community perceptions of co-existence with elephants for a separate element of this overall

study.

Additionally, between January 2015 and December 2017, 15 camera traps were mounted at strategic

locations on posts or trees within the village, and at the forest periphery, to capture footage of

elephants with the aim of determining sex, group size and individual identity. Due to accessibility,

water/elephant damage and availability of charged batteries, the number of camera traps operational

at any given time varied between seven and 15.


To address the questions of whether certain factors influenced a) the likelihood of elephants entering

Dewagiriya Village, b) within-village variations in frequency of elephant events, and c) the likelihood

of elephant-inflicted damage occurring, I selected predictor variables for each based on

environmental, temporal, spatial or social conditions deemed likely to influence elephant behaviour

(table 3.1). The relationship of the associated predictor variables to: a) the likelihood of elephants

entering Dewagiriya Village, and c) the likelihood of elephant-inflicted damage occurring was tested

using generalised linear models, assuming a binomial distribution. The relationship of predictor

variables to b) within-village variations in frequency of elephant events, was tested using a generalised

linear mixed model, assuming a poisson distribution.

Analysis for the datasets relating to each of the above questions followed the same basic procedure.

First, I checked for collinearity of predictor variables using Spearman’s Rank Correlation, as correlated

93

predictors can produce unreliable parameter estimates (Symonds and Moussalli 2011; Quinn and

Keough 2002). All correlation coefficients were <0.40 indicating that no strong correlations existed,

thus all predictor variables were included in the modelling process. For all combinations of models, an

information theoretic approach was employed (Burnham and Andersen 2002), specifically Akaike’s

Information Criterion, adjusted for small sample sizes (AICc). Akaike weights (wi) were calculated to

gauge the level of support for each model, given the sample data (Burnham and Andersen 2002). AICc

differences (∆ i), which show the variation between the highest ranked models and other models in

the candidate set, were calculated to measure support for each candidate model. Models with ∆i <2

were considered to have substantial support (Burnham and Andersen 2002) and predictor variables

were considered influential when the 95% confidence interval (CI) of their coefficient did not overlap

zero.


Generalised linear models (GLM’s) and generalised linear mixed models (GLMM’s) were constructed

using the package ‘lme4’ (Bates et al. 2014), with model outputs generated with the ‘effects’ package

(Fox 2003) and visualised with ‘ggplot2’ (Wickham 2016). Models were ranked and assessed for ‘best

fit’ using the ‘AICcmodavg’ package (Mazerolle 2016).

3.4 Results

3.4.1 Elephant events by raid days

During the three-year study period, one or more elephants were observed in Dewagiriya Village on

293 (25.7%) of 1096 possible raid days. Elephant events occurred almost exclusively at night,

specifically between six pm and three am, with the heaviest traffic occurring between nine pm and

three am. Elephant sightings during daylight hours were rare, comprising only 5.8% of all elephant

events. Predominantly, elephants entered the village in small groups, with a mean (±SE) group size of

1.89 (±0.06) elephants. The maximum number of elephants observed together was eight however

groups of more than four individuals were rarely observed (<4% of elephant events). Lone elephants

were the most common accounting for 47.1% of elephant events.

94

Given the primarily nocturnal nature of elephant events, group type and sex could be difficult to

ascertain. At times people were unaware of elephant presence until evidence was seen the next

morning, and some people were simply unable to differentiate between the sexes. Understandably,

people were more focused on deterring elephants from their property than determining sex and group

type. Still, for 52.2% of events reported, the participant confidently recorded group type. Results

supported a strong male sex bias in elephant activity within Dewagiriya Village with just over 50% of

elephant events attributed to lone males (36.9%) or small all-bull groups (13.3%) and only 2%

attributed to mixed-sex groups. No female-led family groups, or lone females, were identified.

Although camera trap footage represented only a snapshot of total elephant events, further support

was lent to the existence of male sex bias in elephant events. Of the 50 times elephants were captured

on camera, 82% showed bull groups of between one to five individuals, and the remaining footage

was of insufficient quality to determine sex (Figure 3.3)

95

.

Figure 3.3 Camera trap footage of elephants in Dewagiriya Village

Top: An adult male elephant forages in the forest on the periphery of Dewagiriya Village.

Bottom: An adult male elephant walks through Mr. Sobana’s field.

96

3.4.2 Factors influencing elephant presence in Dewagiriya Village

I selected month, year, moon phase, raid day rain, and previous month rain as likely predictors

influencing elephant events in Dewagiriya Village. Specifically, I hypothesised that: yearly and monthly

fluctuations in frequency of elephant events would be observed; that elephant events would peak

during the new moon phase when nights are at their darkest; and that elephant events would increase

when the previous month rain was high but that on nights of actual rainfall, elephants would enter

the village less frequently.

Using data collected on dates of elephant events plus the randomly generated no-elephant event

dates, I developed and compared (using AICc) a total of 22 models encompassing all possible

combinations of the predictor variables. The resulting model selection was four models with Δ i <2,

comprising the 95% confidence interval (CI) set (Table 3.2). Model fit was low for all models (D²≤0.10).

Adopting a backwards (step-down) selection process (Zuur et al. 2009), I fitted each of these models

separately and selected all influential variables (those whose 95% CI did not overlap 0) to include in

the final ‘best’ model from which to draw influence. Results indicated that three variables were

important predictors of elephant events: month, year and raid day rain.

Candidate model df LogLik AICc Δi wi cum. wi D2

Intercept + Raid day rain + Month 14 -361.51 751.82 0.00 0.36 0.36 0.10

Intercept + Month + Year 15 -360.91 752.74 0.92 0.23 0.59 0.10

Intercept + Previous month rain +

Raid day rain + Month

15 -361.08 753.09 1.27 0.19 0.78 0.10

Moon phase + Raid day rain +

Month

17 -359.22 753.62 1.80 0.15 0.93 0.09

Table. 3.1: Results from models predicting likelihood of elephant events in Dewagiriya Village.

All models that gained substantial support (Δi AICc <2 and/or cumulative wi = ≤ 0.95) are included. Variables

are described in methods. Also listed are degrees of freedom (df), LogLikelihood (LogLik), AIC values (AICc),

Delta AICc (Δi), Akaike weight (wi), cumulative Akaike weight (cum. wi) and deviance explained (D2) for each

model.

97

Elephants were observed in Dewagiriya Village during all months of the year however fluctuations

were evident. The ‘best’ model predicted that elephant events would be most frequent during April

and November, then relatively constant for all other months except January and October when a

significant reduction in elephant events would occur (Figure 3.4). A substantial decline in elephant

events per year was also evident, with significantly less elephant events occurring in 2017 (p=0.004)

(Figure 3.5). Finally, as hypothesised, raid day rain impacted elephant activity with elephants

significantly less likely to enter Dewagiriya Village on days when rain fell (Figure 3.6).

It must be noted however, that although all three of these predictor variables showed significant

influence on elephant events, only 11% of the overall variation was explained by the best model (R² =

0.112). Neither the previous month rain nor moon phase significantly influenced the likelihood of

elephant events in Dewagiriya Village.

Figure 3.4 Elephant events per month based on ‘best’ model predictions

98

Figure 3.5. Elephant events per year based on ‘best’ model predictions

99

Figure 3.6 Relationship between raid day rain and elephant events based on ‘best’ model predictions.

100

3.4.3 Factors influencing within-site variations in elephant events

A total of 437 elephant events were reported by individual households. Frequency of elephant events

varied widely per household: at one end of the spectrum only two elephant events were reported by

the Rathnasena family (H31) yet at the other end, the Sumanawathi family (H35) sighted elephants on

their property on 63 occasions. Overall, the mean (± SE) count of elephant events over the entire study

period per household was 16.18 (±2.89).

I selected location in village (north/south), proximity to road, total number of adults in household, and

mean annual income as likely predictors influencing the frequency of elephant visits per household in

Dewagiriya Village. Specifically, I hypothesised that: participants who lived in the south of the village

and whose property did not border the main road would be subject to more frequent elephant events;

as would households with fewer adult members and lower mean annual income, as these households

would likely be less well-equipped to invest in elephant deterrents.

To account for all possible combinations of the predictor variables, 15 models were built and

compared. The top-ranked models comprised two models with Δ i <2, and eight within the full 95% CI

set (Table 3.3). I fitted each of these models in the 95% confidence set separately and selected all

influential variables to include in the final ‘best’ model. The best model indicated that only location

and proximity to road significantly influenced the frequency of elephant events per household.

Households located to the south of the main road in Dewagiriya Village were visited by elephants

significantly more often than households located to the north. Participants whose properties bordered

the main road were significantly less likely to experience high levels of elephant events. This is likely

because properties in the south of the village, and particularly those set back from the main road,

were in closest proximity to the water tanks and forest habitat, which are both valuable resources to

elephants. Overall, results showed 46.4% of the variance in frequency of elephant visits to explained

by the ‘best’ model when only fixed effects were considered (R₂(marginal) = 0.463), and 80% of the

variance was explained when fixed and random effects were accounted for (R₂(conditional) = 0.798).

101

3.4.4 Damage inflicted by elephants

Of the 293 raid days where elephant events were documented, damage occurred 71.7% of the time.

The intensity of damage varied considerably falling anywhere in between a single fruit tree being

raided, to a full two acres of rice crops destroyed, to multiple walls of a home being knocked down

(Figure 3.7).

Candidate model df LogLik AICc Δi wi cum. wi R₂ (m) R₂ (c)

Intercept + Location +

Proximity to road

4 -197.19 403.00 0.00 0.34 0.34 0.44 0.80


Proximity to road + Income

5 -196.59 404.13 1.12 0.19 0.54 0.46 0.80


Proximity to road + # adults

5 -197.09 405.13 2.12 0.12 0.65 0.44 0.80

Intercept + Proximity to road

+ # adults

4 -198.60 405.83 2.83 0.08 0.74 0.38 0.79


+ income

6 -196.38 406.11 3.11 0.07 0.81 0.47 0.80


+ # adults + income

5 -197.70 406.35 3.35 0.06 0.87 0.41 0.79

Intercept + Proximity to road 3 -200.48 407.34 4.33 0.04 0.91 0.30 0.81


+ # adults

4 -199.66 407.94 4.93 0.03 0.94 0.33 0.79

Table 3.3: Results from models predicting within-site variations in frequency of elephant events per household

All models that gained substantial support (Δi AICc <2 and/or cumulative wi = ≤ 0.95) are included. Variables are

described in methods. Also listed are degrees of freedom (df), LogLikelihood (LogLik), AIC values (AICc), Delta

AICc (Δi), Akaike weight (wi), cumulative Akaike weight (cum. wi) and deviance explained (R₂ (marginal) and R₂

(conditional)) for each model.

102

Figure 3.7 Elephant damage in Dewagiriya Village. Top Left: elephant footprint damage in the rice field. Top

Right: Elephants visited the Thilanarathne family during the night, knocking down part of the wall while they

were inside. Bottom: elephant damage to banana trees at the Ghanawathi home garden.

103

Specific crops grown varied according to household and year, however between 2015 and 2017 all

participants farmed rice, and more than 75% of participants also planted corn (22/27), mung beans

(22/27), miscellaneous vegetables (one or more of: avocado, beans, eggplant, okra, pumpkin, sweet

potato and bitter gourd) (22/27) and bananas (25/27). Of these, rice and bananas were the most

frequently damaged, on 25.6% and 28% of days when elephant events occurred respectively. Coconut

trees and manioc were the next most frequently damaged plants (13% and 11.6% of raid days

respectively). 70.4% of households had coconut trees on their property, and 59.3% grew manioc

periodically. Several plants were subjected to high incident of damage relative to the number of

households who grew them, with >80% of farmers who grew sugar cane, chili or millet experiencing

damage to the plant in question, at least once during the study period (Table 3.4). 37% of participants

also experienced damage to their home at least once during the study period.

Crop/plant type % times

damaged

(n = 293)

% households

growing

(n = 27)

% households with

elephant damage

Banana 28% 92.6% 88% (n=25)

Cashew 2% 22.2% 33.3% (n = 6)

Chili 4.1% 40.7% 90.9% (n = 11)

Citrus (limes and oranges) 0% 25.9% 0% (n = 7)

Coconut 13% 70.4% 79% (n = 19)

Corn 3.1% 81.5% 22.73% (n = 22)

Cow peas 0.4% 55.6% 6.67% (n = 15)

Manioc 11.6% 59.3% 93.8% (n=16)

Millet 5.8% 51.9% 85.71% (n = 14)

Mung beans 4.4% 77.8% 38.1% (n = 21)

Other fruit (guava, jackfruit,

papaya, mango, watermelon,

tamarind)

5% 70.4% 47.4% (n = 19)

Other vegetables (including

beans, eggplant, okra, pumpkin,

sweet potato, bitter gourd)

5.12% 81.5% 27.3% (n = 22)

Peanuts 2% 25.9% 14.29% (n = 7)

Rice 25.6% 100% 70.37% (n = 27)

Sugar cane 1% 22.2% 100% (n = 6)

Table 3.4 Crop and plant types grown and damaged in Dewagiriya Village. Bold text denotes: crops most frequently

damaged (per crop type and per household) and most frequently grown

The 2nd column shows how often each crop/plant type was damaged as a % of total raid days. Bold text represents the 4 most commonly damaged. The 3rd column shows how many of the participating households grew each crop/plant at least once between 2015 and 2017. The 4th column shows the % of households who experienced damage to the given crop/plant at least once between 2015 and 2017 (excluding households who did not grow the crop/plant). For the 3rd and 4th columns, bold text represents scores of ≥75%).

104

As with presence or absence of elephant events, I hypothesised that certain factors would influence

whether damage was inflicted when elephant events occurred. Again, I selected month, year, moon

phase, raid day rain, and previous month rain as my predictor variables, expecting monthly and annual

fluctuations in damage frequency, more damage to occur during new moon phases and when the

previous months rainfall was high, and less frequent damage on nights of actual rainfall.

To test for possible relationships between the predictor variables and presence/absence of damage

occurring on nights of elephant events, I incorporated all possible combinations of the predictor

variables to test 22 models. The resulting model selection comprised two models with Δ i <2. These

models, along with four others, comprised the 95% CI set (Table 3.5). I fitted each of these models

separately and selected all influential variables to include in the final ‘best’ model.

Results indicated that previous month rain and year were important predictors of elephant damage.

Specifically, elephant damage increased in frequency when the previous month rain was high but

decreased significantly in 2017 compared to other years. These results were represented in both the

prediction models and observed data and are likely a reflection of increased crop growth following

high rainfall, and reduced crop availability in Dewagiriya Village in 2017 due to poor availability of

water. Still, only 4.2% of the overall variation was explained by this ‘best’ model (R² = 0.042).

Candidate model df LogLik AICc Δi wi cum. wi D2

Intercept + Prev month rain + Year 4 -150.31 308.78 0.00 0.43 0.43 0.04

Intercept + Prev month rain + Raid day

rain + Year

5 -150.02 310.26 1.48 0.21 0.64 0.04

Intercept + Year 3 -152.82 311.73 2.95 0.10 0.74 0.03

Intercept + Prev month rain 2 -153.94 311.92 3.14 0.09 0.83 0.02

Intercept + Raid day rain + Year 4 -151.94 312.04 3.26 0.09 0.91 0.03

Intercept + Prev month rain + Raid day

rain

3 -153.71 313.52 4.74 0.04 0.95 0.02

Table 3.5 Results from models predicting likelihood of damage during elephant events in Dewagiriya Village.

All models that gained substantial support (Δi AICc <2 and/or cumulative wi = ≤ 0.95) are included. Variables are

described in methods. Also listed are degrees of freedom (df), LogLikelihood (LogLik), AIC values (AICc), Delta

AICc (Δi), Akaike weight (wi), cumulative Akaike weight (cum. wi) and deviance explained (D2) for each model.

105

3.5 Discussion

Overall, elephant events in Dewagiriya Village were found to be widespread and frequent, with all

participants suffering negative impacts of crop-raiding, to varying degrees. How levels of elephant

activity in Dewagiriya Village compare with other known HEC locations was difficult to ascertain due

to differences in data recording (elephant events per raid day vs. individual elephant incidents per

household, vs. damage incidents only), how many households within an area of interest were

surveyed, and delineation of study site boundaries (human inhabited areas only or inclusion of

bordering PA’s in measurements). Of 57 peer-reviewed publications identifying crop-raiding patterns

and influences throughout African and Asian elephant home ranges, only six included the information

required to directly compare a calculable rate of crop-raiding. Dewagiriya Village’s annual rate of 48.8

elephant events per km² was exceptionally high relative to other locations (Table 3.6). While this does

not suggest that the crop-raiding problems experienced elsewhere are not serious, it does highlight

the extreme situation faced by the Dewagiriya community. A more uniform global approach to

quantifying elephant events would be useful in facilitating meaningful comparisons within and

between elephant ranges and assist with identifying priorities for mitigation action.

Country Location Size of

study site

(km²)

Duration

of study

(months)

#

incidents

Rate of

CR per

km² (pa)

Reference

China Yunnan Province 19582 60 18261 0.19 Chen et al. 2005

Ghana Kakum CA

10 26 300 13.84 Barnes et al. 2007

India Assam 200 36 390 0.65 Das et al. 2012

India Assam 1250 36 1761 0.47 Davies et al. 2011

Sri

Lanka

South-east Dry

Zone

Not

provided

12 383 2.2 Ekanayaka et al.

2011

Sri

Lanka

South-east Dry

Zone

262.2 12 975 3.72 Campos-Arceiz et

al. 2009

Sri

Lanka

Central Dry Zone 2 36 293 48.8

Table 3.6. Comparison of annual crop-raiding counts per km² across HEC locations. Bold text denotes results

from this study

106

3.5.1 Characteristics of crop-raiding elephants

In concurrence with common characteristics of elephants at most crop-raided locations (see: Africa:

Graham et al. 2010; Sitati et al. 2003, Asia: Aziz et al. 2016; Borah and Bhuyan 2016; Thaufeek et al.

2014; Das et al. 2012; Haturusinghe and Weerakoon 2012; Nath et al. 2013; Campos-Arceiz et al.

2009), it was found that solitary elephants or small groups most commonly utilised crop-raiding as a

foraging strategy in Dewagiriya Village. Globally, it is typical for male elephants to dominate crop-

raiding activity (Africa: Chiyo et al. 2012; Graham et al. 2010, Asia: Borah and Bhuyan 2016; Nath et

al. 2013; Das et al. 2012; Haturusinghe and Weerakoon 2012; Ekanayake et al. 2011; Fernando et al.

2005; Wikramanayake et al. 2003; Sukumar 1988), and again, results from Dewagiriya appear to

concur with this general trend. Of 155 events where sex was determined, on only six occasions were

females involved, indicating a strong male-bias. This is further supported by camera trap evidence of

nine individually identified males in or on the periphery of Dewagiriya, all bearing wounds and scars

consistent with human inflicted injuries (pers. comm). It should be noted though, that for 47% of

elephant events, community members could not accurately identify sex, and that in general people

were more confident identifying adult males than differentiating between younger male and adult

female elephants. Thus, it is possible that more females than reported were involved in crop-raids,

and female-led herds are known to raid intermittently in areas of south-east (Fernando et al. 2005),

north-central (Wikramanayake et al. 2003), and north-west Sri Lanka (Haturusinghe and Weerakoon

2012). Still, it seems unlikely that raiding by females is anything bar a rare event in Dewagiriya Village,

although further research using structured night-time observations and/or mitochondrial DNA

analysis on elephant faeces would be beneficial for confirmation.

It has been suggested that male and female elephants differ in their propensity to take risks, and that

crop-raiding is one such behaviour where this difference is apparent (Ekanayaka et al. 2011; Sukumar

and Gadgil 1988). Despite having equal access to cultivated fields, male elephants are typically more

likely to adopt crop-raiding as a high-risk, high-gain strategy, benefitting from bountiful supplies of

nutritionally high foods, while females prefer to forage in the relative safety of wild habitat rather

than exposing themselves and their young to the potential risk of human interaction (de Silva et al.

2013; Haturusinghe and Weerakoon 2012; Fernando et al. 2011). Furthermore, studies in Asia

(Ekanayaka et al. 2011) found that non-raiding female herds obtained ample resources from wild

fodder without needing to delve into cultivated fields as some of their male counterparts were doing,

107

indicating that neither nutritional benefits nor access fully explain the motivation to crop-raid.

Although outside the scope of this study, I saw no obvious distinction between body condition of adult

females and crop-raiding males, or indication that ample nutrition could not be obtained without

raiding crops (K. Butler, pers. obs). Further research combining an assessment of body condition and

nutritional intake, with detailed behavioural studies of daily activities of crop-raiding and non-crop-

raiding individuals, would go a long way to understanding the motivational states of crop-raiding

elephants, and perhaps expose another layer of characteristics and influences driving crop-raiding

behaviour. However, this requires a long-term commitment and is fraught with difficulties in forested

habitat, so in the meantime mitigation planning based on current evidence and more efficient

assessments of crop-raiding activity must go on.

3.5.2 Influences of crop-raiding behaviour

Investigating broad influences of crop-raiding behaviour in Dewagiriya Village, including timing of

raids, targeted crop types, and seasonal variations, I found no wild deviations from the norms

identified throughout both Asia and Africa. In all locations where crop-raiding is a major HEC issue,

elephants utilise this foraging strategy nocturnally, and it is extremely rare to see elephants in villages

during the day (Wilkie and Douglas-Hamilton 2018; Das et al. 2014; von Gerhardt et al. 2014; Nath et

al. 2013; Graham et al. 2010; Santaipillai et al. 2010; Campos-Arceiz et al. 2009; Jackson et al. 2008;

Sitati et al. 2003). Unsurprisingly, Dewagiriya Village is no exception and although very occasionally

elephants were seen at water tanks in the late afternoon, only once throughout the three-year study

did an elephant enter a field during daylight. Furthermore, elephant crop-raiding occurred most

frequently between 9 pm and 3 am, the hours of darkness where outdoor human activity is virtually

non-existent. Although most studies do not specify times of raiding beyond ‘nocturnal’, similar peaks

in specific raiding times when people are generally inside and asleep have been identified in Kenya

(Sitati et al. 2003) and Sri Lanka (Santiapillai et al. 2010).

‘Awareness’ and ‘forward-thinking’ in animals are difficult to prove conclusively, but still many

prominent researchers believe elephants possess an awareness of, and ability to assess, the potential

danger posed by interactions with humans (Wilkie and Douglas-Hamilton 2019; Gunn et al. 2013;

Graham et al. 2009; Barnes et al. 2006). Foraging on crops under the dark cover of night, and

specifically adjusting nocturnal foraging to hours of lowest human activity, is likely a risk-avoidance

108

strategy that some elephants adopt to minimise the risk of human encounters, while still benefiting

from access to highly nutritious cultivated foods (Gunn et al. 2013; Graham et al. 2009; Barnes et al.

2006). Patterns identified in Dewagiriya Village, coupled with farmer reports and anecdotal evidence

from camera trap footage of elephants resting and foraging in the peripheral forest during the day,

concur with these risk-minimisation theories.

In Ghana (Barnes 2007) and Tanzania (Gunn et al. 2013) it appears elephants have taken risk-

minimisation strategies one step further. Not only do elephants raid nocturnally, they are also

influenced by the light availability of the moon, and crop-raid more frequently during new lunar

phases when nights are at their darkest. This would presumably reduce the likelihood of being

detected by humans even further. In Dewagiriya Village however, moon phase was not found to be

an influencing factor on crop-raiding activity. Perhaps the extremely high frequency of elephant events

precluded influence of moonlight, or perhaps other factors were simply more influential.

It was interesting to note that in Dewagiriya Village, elephants were less likely to crop-raid on nights

when rain fell. I could not find any other studies considering the effects of daily rainfall on crop-raiding,

so it is impossible to know if this is typical or atypical. I can only hypothesise that elephants utilising

habitat in and around Dewagiriya Village, preferred to remain under the relative shelter of the forest

during rain. While certainly not the be-all and end-all of understanding crop-raiding patterns,

identifying very specific influences of crop-raiding behaviour such as lunar phase or nightly rain, could

at least help farmers plan mitigation by knowing when crop-guarding efforts need to be intensified or

could reasonably be relaxed.

In most locations, the crop most targeted by elephants is the crop most frequently grown.

Unfortunately, this generally means the crop that is most valued by farmers, for subsistence or income

generation or both. In many areas of Asia this crop is rice (also called paddy) (e.g. Patil and Patil 2017;

Borah and Bhuyan 2016; Palei and Singh 2016; Nath et al. 2013; Sahu and Das 2012; Davies et al. 2011;

Ekanayaka et al. 2011; Webber et al. 2011; Santiapillai et al. 2010; Sarkar and Roskaft 2010; Campos-

Arceiz et al. 2009), while in Africa maize is often the most heavily impacted (e.g. Amwata and Mganga

2014; Guerbois et al. 2012; Graham et al. 2010; Sitati and Walpole 2006; Chiyo et al. 2005). Commonly

grown fruits, such as banana (Africa and Asia) or coconut (Asia only), are also highly susceptible to

elephant raids (e.g. bananas: Berliani et al. 2016; Chen et al. 2016; Harich et al. 2013; Nath et al. 2013;

109

Sahu and Das 2012; Ekanayaka et al. 2011; Webber et al. 2011; Chiyo et al. 2005; Sitati et al. 2003,

bananas and coconut: Patil and Patil 2017; Borah and Bhuyan 2016; Palei and Singh 2016; Santaipillai

et al. 2010; Campos-Arceiz et al. 2009). Dewagiriya Village was no different, with rice and bananas the

most frequently damaged crops. This is no surprise given that rice was grown annually by all

participants and bi-annually by farmers who had year-round access to water, and bananas were a

common feature in all home gardens. Coconut trees, also grown by almost every family, and both

manioc and millet which some farmers grow as an alternative to rice during the ‘short rains’, were

also frequently raided. Overall, 25 cultivated crop types were identified as being damaged by

elephants in Dewagiriya Village, including trampling of unpalatable crops such as chili.

Admittedly, the assessment of crop types damaged in this study is simplistic, fulfilling only the aim of

identifying which crops were damaged most frequently. Given the extremely high level of elephant

activity and the variety of crops grown, collecting detailed information on timing of plant growth,

proportion of crop damage relative to plot size, and identifying economic losses based on growth stage

of plants, was beyond the capacity of this study. However, there is no denying that this information

would be valuable in determining precise economic losses caused by elephant presence for the overall

community, and individual households, and giving further depth to impact assessments. It should also

be noted that some households in Dewagiriya, refrain from growing certain crops that were previously

valuable to them, due to the risk of elephant depredation.

A strategy often put forward to mitigate human-elephant conflict is to change cultivation types (e.g.

Berliani et al. 2016; Palei and Singh 2016; Allendorf et al. 2015), or to plant buffer zones of unpalatable

crops (e.g. Gross et al. 2017; Allendorf et al. 2015; Gross et al. 2015; Chiyo et al. 2005). For

communities, such as Dewagiriya, where the main crop is the one relied upon for subsistence, a major

shift in cultivation type is highly unlikely to be practical or supported by the community. In Dewagiriya,

planting buffer zones of unpalatable crops would also be difficult, given that rice fields and gardens

are already relatively small, and the risk of trampling would be high. Still, including small plots of high

value unpalatable crops close to the house or within small fenced areas might be worthwhile, as it

could both reduce the proportion of crops in the home garden attractive to elephants and generate

income through market sales. Several types of medicinal and aromatic plants have already proved

promising as small-scale alternative crops in both Africa (Gross et al. 2015) and Asia (Gross et al. 2017).

It would be well worth investigating options to help rural communities on a site-specific basis.

110

However when designing trials or proposing changes, it needs to be remembered that in villages such

as Dewagiriya, most families have very little disposable income, and people would either need to be

supported through initial trials, or scientific evidence of success should be strong before asking

farmers to invest their own money in planting alternative unpalatable crops.

Finally, although elephants were present in Dewagiriya all throughout the year, notable peaks and

troughs in activity existed. In most locations, such fluctuations are influenced by seasonal rainfall and

subsequent crop growth and harvests (Borah and Bhuyan 2016; Djagoun et al. 2016; Palei and Singh

2016; Das et al. 2014; Nath et al. 2013; Das et al. 2012; Gubbi 2012; Haturusinghe and Weerakoon

2012; Santiapillai et al. 2010; Campos-Arceiz et al. 2009; Jackson et al. 2008; Sitati and Walpole 2006;

Chiyo et al. 2005; Sukumar 1990; Sukumar 1988), however in Dewagiriya Village there was no clear

influence of rainfall or season on crop-raiding. Elephant events were typically low during January and

February despite this being part of the ‘long rain’ season, with paddy fields abundant and on their way

to maturity. It could be that abundance of natural fodder outside of human habitations during this

time, or increased crop-guarding by farmers as rice fields begin to flourish, impacted a reduction in

elephant visits. However, these factors were not investigated during the study and are therefore

speculations, warranting further investigation. Elephant events were also low in October, which is

before the commencement of the ‘long rains’ and towards the end of the dry season when only the

hardiest of plants grow. The lack of clear patterns relating to rainfall or season was likely a product of

unusual weather, with rainfall in 2016 and early 2017 well below average. As a result, many farmers

altered their planting patterns, with some forgoing planting of rice, manioc or millet during the ‘short

rains’ and many postponing and/or planting smaller plots of rice than usual during the ‘long rains’. To

confidently determine if influences of rainfall or season impact elephant activity in Dewagiriya Village,

on-going long-term monitoring of elephant events measured against rainfall and crop-growth are

recommended.

3.5.3 Within-site variations in elephant events

Developing an understanding of the characteristics and influences driving crop-raiding at a particular

village or location, is a useful tool for facilitating meaningful comparisons across HEC sites, for

developing mitigation techniques, and for identifying times or circumstances where crop-guarding

efforts need to be most concentrated. Still, crop-raiding is rarely uniform throughout a village, and

111

identifying within-site nuances and their potential influences, should also be a priority in crop-raiding

studies. This can help to determine within-village hotspots of elephant activity where mitigation

efforts should be focused, or to identify families requiring additional support.

In Dewagiriya Village, households located to the south of the ‘main road’ (a 2.5 metre wide dirt track

running east to west through the village) were much more susceptible to crop raids then those to the

north, and this was further amplified for families whose properties did not border the road. The main

surrounding forest habitat for elephants, plus four large water tanks, are all located to the south of

Dewagiriya, therefore it makes sense that those closest to elephant refuge bear the brunt of elephant

activity. Houses located closer to the road are to some degree buffered by properties behind their

own and closer to the village edge, meaning that there is more chance that elephants will be detected

by their neighbours before reaching those properties. This ‘buffer’ of properties between the forest

and fields and homes bordering the road, is probably more influential than the road itself, although

the increased human activity as people traverse the village using the road, coupled with a slightly

higher density of dwellings may also contribute to reduced elephant activity.

This proximity to elephant habitat or refuge as a predictor of crop-raiding incidence, holds sway in

both large-scale studies investigating elephant events over multiple villages (e.g. Santiapillai et al.

2010) and at the micro-scale. Although only a few studies have investigated specific within-site

variations, an increased vulnerability of households on the outskirts of villages is consistent, not only

when considering crop-raiding by elephants (REF) but for other species involved in crop-raiding such

as baboons (Hill 2000). This goes back to the ‘risk-minimisation’ theory, as logically elephants

entertain a somewhat smaller risk by raiding in fields closest to their escape route (back to the forest),

then venturing into the midst of a village where they are likely to encounter more people. Even in a

very small village such as Dewagiriya, where distances between properties on the periphery and

properties adjacent to the main road can be as little as 500 m, and only a narrow road divides the

north and south, there was an obvious increase in elephant events closest to the forest.

It is supposed that socio-economic factors can be a predictor of vulnerability to crop-raiding, however

this was not the case in Dewagiriya Village. Although households with less annual income might be

less equipped to invest in elephant deterrents, they are also less likely to have multiple rice growing

seasons per year, and typically have smaller crop plots and home gardens, so less area for elephants

112

to damage. Furthermore, I hypothesised that households with less adults might be more susceptible

to crop-raids, having less manpower to defend their crops but this was not the case, perhaps because

neighbours often join together to guard against elephants, especially if occupants of the household

are elderly or a single adult living alone.

3.6 Conclusion

This study of localised human-elephant conflict in a rural Sri Lankan village, provides a three-year

snapshot of the frequent crop-raiding experienced by a small farming community heavily reliant on

crop production and home-grown vegetables for income generation and subsistence. Previously

unquantified, we now have a clearer idea of how elephant activity in Dewagiriya Village fits into the

bigger picture of HEC in Sri Lanka, and indeed other locations of Asia and Africa experiencing similar

issues. Ultimately, the similarities across sites (e.g. nocturnal foraging strategy primarily undertaken

by males in small groups and targeting commonly grown crops) are greater than the differences (e.g.

extremely high frequency of elephant events and no clear seasonal patterns), although this is not to

downplay the importance of identifying site-specific nuances to develop appropriate mitigation

strategies.

Overall, our data show that elephant crop-raiding occurs year-round and that although farmers would

be wise to increase crop-guarding efforts when their most valuable crops are at stake, there is no time

of year when they can afford to lapse with vigilance. As elephants raid more often on nights of no

rainfall, it would pay to increase guarding efforts in clear conditions. If communal guarding or

defending of crops was considered, concentrating mitigation efforts in the south area of the village,

and particularly along the fields bordering the water tanks and forest should be of highest priority, as

preventing elephant entry here would likely benefit most of the community. Further study specifically

investigating a variety of appropriate mitigation techniques is essential to help this community.

Importantly, data collected during studies such as this should contribute not only to the greater

discussion of human-elephant conflict and its drivers and consequences but should be of practical use

to the communities involved. It is hoped that this systematically produced profile of crop-raiding in

Dewagiriya, can be used by the community to demonstrate their plight and need for assistance to local

authorities, and also within the community to assist with mitigation planning.

113

3.7 References

Barnes, RFW, Hema, EM, Nandjui, A, Manford, M, Dubuire, UF, Danquah, EKA & Boafo, Y 2005, 'Risk

of crop-raiding by elephants around the Kakum Conservation Area, Ghana', Pachyderm, vol. 39, pp.

19-25.

Barnes, RFW, Dubiure, UF, Danquah, E, Boafo, Y, Nandjui, A, Hema, EM & Manford, M 2006, 'Crop-

raiding elephants and the moon', African Journal of Ecology, vol. 45, pp. 112-115.

Baruch-Mordo, S, Breck, SW, Wilson, KR & Theobald, DM 2008, 'Spatiotemporal distribution of black

bear-human conflicts in Colorado, USA', Journal of Wildlife Management, vol. 72 (8), pp.1853 – 1862.

Campos-Arceiz, A, Takatsuki, S, Ekanayaka SKK & Hasegawa, T 2009, ‘The human-elephant conflict in

Southeastern Sri Lanka: Type of damage, seasonal patterns, and sexual differences in the raiding

behaviour of elephants’, Gajah, vol. 31, pp. 5 – 14.



Environmental Development and Sustainability, vol. 15, pp. 547-565.

Chen, Y, Marino, J, Chen, Y, Tao, Q, Sullivan, CD, Shi, K & MacDonald, DW 2016, 'Predicting hotspots

of human-elephant conflict to inform mitigation strategies in Xishuangbanna, Southwest China', PLoS

ONE, vol. 11(9): e0162035.doi:10.1371/journal.pone.0162035.

Choudhury, A 2004, 'Human-elephant conflicts in Northeast India', Human Dimensions of Wildlife, vol.

9(4), pp. 261-270.

Das, BJ, Lahkar, BP & Talukdar, BK 2012, 'Increasing trend of human elephant conflict in Golaghat

District, Assam, India: Issues and concerns', Gajah, vol. 37, pp. 34-37.


and crop-raiding by wild Asian elephant (Elephas maximus) in Sonitpur District of Assam,

114

Davies, TE, Wilson, S, Hazarika, N, Chakraborty, J, Das,D, Hodgson, DJ & Zimmerman, A 2011,

'Effectiveness of intervention methods against crop-raiding elephants', Conservation Letters, doi:

10.1111/j.1755-263X.2011.00182.x

Dickman, AJ 2010, ‘Complexities of conflict: the importance of considering social factors for effectively

resolving human-wildlife conflict’, Animal Conservation, vol. 12, pp. 459 – 466.

de Silva, M & de Silva, PK 2007, The Sri Lankan elephant: Its evolution, ecology and conservation, WHT

Publications, Colombo.

Ekanayaka, SSK, Campos-Arceiz, A, Rupasinghe, M, Pastorini, J & Fernando, P 2011, 'Patterns of crop-


vol. 20-25.

Fernando, P, Wikramanayake, E, Weerakoon, D, Jayasinghe, LKA, Gunawardene, M & Janaka HK 2005,

‘Perceptions and patterns of human-elephant conflict in old and new settlements in Sri Lanka: insights

for mitigation and management’, Biodiversity and Conservation, vol. 14, pp. 2465 – 2481.



Graham, MD, Douglas-Hamilton, I, Adams, WM & Lee, PC 2009, 'The movement of African elephants

in a human-dominated land-use mosaic', Animal Conservation, vol. 12, pp. 445-455.

Graham, MD, Notter, B, Adams, WM, Lee, PC & Ochieng, TN 2010, ‘Patterns of crop-raiding by

elephants, Loxodonta Africana, in Laikipia, Kenya, and the management of human-elephant conflict’,


Gunn, J, Hawkins, D, Barnes, RFW, Mofulu, F, Grant, RA & Norton, GW 2013, 'The influence of lunar

cycles on crop-raiding elephants: evidence for risk avoidance', African Journal of Ecology, vol. 52, pp.

129-137.

115

Harisch, FK, Treydte, AC, Sauerborn, J & Owusu, EH 2013, ‘People and wildlife: conflicts arising around

the Bia Conservation Area in Ghana’, Journal for Nature Conservation, vol. 21, no. 5, pp. 342 – 249.



Hedges, S, Tyson, MJ, Sitompul, AF, Kinnaird, M, Gunaryadi, D & Aslan 2005, 'Distribution, status and

conservation needs of Asian elephants (Elephas maximus) in Lampung Province, Sumatra, Indonesia',

Biological Conservation, vol. 124, pp. 139-146.

Hill, CM 2004, ‘Farmers perspectives of conflict at the wildlife-agriculture boundary: some lessons

learned from African subsistence farmers’, Human Dimensions of Wildlife, vol. 9, pp. 279 – 286.

Ikanda, D & Packer, C 2008, 'Ritual vs. retaliatory killing of African lions in Ngorongoro conservation

area, Tanzania', Endangered Species Resources, vol. 6, pp. 67-74.

Kansky, R, Kidd, M & Knight, AT 2016, 'A wildlife tolerance model and case study for understanding

human wildlife conflicts', Biological Conservation, vol. 201, pp. 137-145.

Leimgruber, P, Gagnon, JB, Wemmer, C, Kelly, DS, Songer, MA & Selig, ER 2003, ‘Fragmentation of

Asia’s remaining wildlands: implications for Asian elephant conservation’, Animal Conservation, vol. 6,

pp. 347 – 359.

Loe, J & Roskaft, E 2004, 'Large carnivores and human safety – a review', Ambio, vol. 33 (6), pp. 283 –

288.

Naughton-Treves, L, Mena, JL, Treves, A, Alvarez, N & Radeloff, VC 2003; Wildlife survival beyond park

boundaries: the impact of slash-and-burn agriculture and hunting on mammals in Tambopata, Peru',

Conservation Biology, vol. 17(4), pp. 1106-1117.

Njiman, V & Nekaris, KA 2010, 'Testing a model for predicting primate crop-raiding using crop-and-

farm-specific risk values', Applied Animal Behaviour Science, vol. 127, pp. 125-129.

116

O'Connell-Rodwell, CE, Rodwell, T, Rice, M & Hart, LA 2000, 'Living with the modern conservation

paradigm: can agricultural communities co-exist with elephants? A five-year case study in East Caprivi,

Namibia', Biological Conservation, vol. 93, pp. 381-391.

Packer, C 2005, 'Lion attacks on humans in Tanzania', Nature, vol. 436 (18), pp. 927 – 928.

Parker, GE, Osborn, FV, Hoare, RE and Niskanen, LS 2007, Human-elephant conflict mitigation: a

training course for community-based approaches in Africa. Participant's manual, Elephant Pepper

Development Trust, Livingstone Zambia and IUCN/SSC AfESG, Nairobi, Kenya.

Perera, BMAO 2009, ‘The Human-Elephant Conflict: A review of current status and mitigation

methods’, Gajah, vol. 30, pp. 41 – 52.

Perez, E & Pacheco, LF 2006, 'Damage by large mammals to subsistence crops within a protected area

in a montane forest of Bolivia', Crop Protection, vol. 25, pp. 933-939.

Puyravaud, JP, Gubbi, S, Poornesha, HC & Davidar, P 2019, ‘Deforestation increases frequency of

incidents with elephants (Elephas maximus)’, Tropical Conservation Science, vol. 12:

https://doi.org/10.1177/194008291986595.

Saraswat, R, Sinha, A & Radhakrishna, S 2015, 'A god becomes a pest? Human-rhesus macaque

interactions in Himachal Pradesh, northern India', European Journal of Wildlife Resources, vol. 61, pp.

435 – 443.

Sitati, NW, Walpole, MJ, Smith, RJ & Leader-Williams, N 2003, ‘Predicting spatial aspects of human-

elephant conflict’, Journal of Applied Ecology, vol. 40, pp. 667 – 677.

Sitati, NW, Walpole, MJ & Leader-Williams, N 2005, 'Factors affecting susceptibility of farms to crop-

raiding by African elephants: using a predictive model to mitigate conflict', Journal of Applied Ecology,

vol. 42, pp. 1175-1182.

117

Sitati, NW & Walpole, MJ 2006, ‘Assessing farm-based measures for mitigation human-elephant

conflict in Transmara District, Kenya’, Oryx, vol. 40, no. 3, pp. 279 – 286.

Studsrod, JE & Wegge, P 1995, 'Park-people relationships: the case of damage caused by park animals

around the Royal Bardia National Park, Nepal', Environmental Conservation, vol. 22(2), pp. 133 – 142.

Sukumar, R 2003, The living elephants: evolutionary ecology, behaviour and conservation, Oxford

University Press, New York.

Sukumar, R 2006, 'A brief review of the status, distribution and biology of wild Asian elephants,

Elephas maximus', International Zoo Yearbook, vol. 40, pp. 1 – 8.










pp. 164-169.

Wilson, ML, Hauser, MD & Wrangham, RW 2007, ‘Chimpanzees (Pan troglodytes) modify grouping

and vocal behaviour in response to location-specific risk’, Behaviour, vol. 144, pp. 1621 – 1653.

Woodroffe, R, Thirgood, S & Rabinowitz, A 2005, 'The impact of human-wildlife conflict on natural

systems', IN: (eds) Woodroffe, R, Thirgood, S & Rabinowitz, A, People and Wildlife: Conflict or

Coexistence? Cambridge University Press, UK.

118

Chapter 4: Beehive fences as a human-elephant conflict mitigation tool for farmers in rural Sri Lanka

119

4.1 Abstract

As crop-raiding continues to threaten both elephant conservation and the livelihoods of farmers living

in proximity to elephants, the need for practical and effective mitigation tools becomes increasingly

important. In recent years, interest in community-based mitigation methods has grown, and there has

been a push towards increased implementation of eco-deterrents such as planting unpalatable crops

or using chili-ropes or briquettes. Beehive fencing is another deterrent showing promise in Africa

however it’s applicability as an Asian elephant deterrent has not been thoroughly explored. In this

study, we initiated a three-year trial, using ten beehive fences to protect home gardens in Dewagiriya,

a small village in Sri Lanka vulnerable to crop-raiding. We found a significant decrease in elephant

incursions into home gardens protected by a beehive fence compared to those without. However,

beekeeping was challenging with colonies frequently absconding and honey harvests low.

Furthermore, despite being relatively inexpensive the costs of establishing a beehive fence were

prohibitive to the Dewagiriya community which is probably reflective of many rural communities in

Sri Lanka. Despite high initial interest in trial participation, farmer motivation fluctuated throughout

the trial, and only 40% of farmers planned to continue using their fences long-term. Further trials in

areas of Asia, preferably with communities already familiar with beekeeping, would be beneficial to

determining its usefulness as a practical, community-based mitigation tool for rural communities

experiencing problems with crop-raiding.

4.2 Introduction

Social, economic and conservation issues resulting from human-elephant conflict (HEC), and

specifically crop-raiding, have been documented in scientific and popular mediums for several

decades (examples: Hoffmeier-Karimi and Schulte 2014; Baskaran et al. 2013; Chakraborty et al. 2013;

Ranasinghe et al. 2010a; 2010b; Bandara and Tisdell 2003; Osborn and Parker 2002; de Silva 1998;

Sukumar 1990). Management plans focusing on deterrent or mitigation methods have been devised,

implemented and assessed with varying levels of success, for various locations (examples: Gunaryadi

et al. 2017; Davies et al. 2011; Perera 2009; Fernando et al. 2008; Graham and Ochieng 2008; Sitati

and Walpole 2006; Dublin and Hoare 2004; Osborn and Parker 2003). However, overall it does not

appear that the problem is alleviating. In fact, the common consensus is that crop-raiding is only

getting worse (Fernando et al. 2011). It is generally accepted that anthropogenically influenced

120

changes to local environments are at the root cause of HEC (Amwata and Mganga 2014; Chakraborty

et al. 2013; Fernando et al. 2011; Campos-Arceiz et al. 2009; Santiapillai et al. 2006; Fernando et al.

2005; Choudhury 2004; Sitati et al. 2003) however the humans who bear the brunt of the problem are

usually not those responsible for the underlying causes (Barua et al. 2013; Santaipillai et al. 2010; Ogra

2008; de Silva 1998). Not surprisingly, communities in HEC zones are increasingly frustrated by the

ongoing challenges presented by living in proximity to wild elephants. Many people feel disillusioned

both by the perceived unwillingness of local authorities to assist (Nsonsi et al. 2018; Mabelunga et al.

2016; Chakraborty and Mondal 2013), and a feeling of helplessness at their own oft-limited capacity

to protect their livelihoods and families (Jasmine et al. 2015; Nath et al. 2015; Bandara and Tisdell

2003).

The question of how best to manage HEC is one with no easy answers. With crop-raiding presenting

challenges in almost every elephant range country, site-specific differences mean there is no one-size-

fits-all solution and most researchers agree that developing a ‘toolbox’ of various deterrents to be

used in combination, or rotated depending on site specific characteristics, would be most appropriate

(van de Water and Matteson 2018; King et al. 2017; Pozo et al. 2017; Hoare 2015; Karidozo and Osborn

2015; Hedges and Gunaryadi et al. 2009; Fernando et al. 2008; Dublin and Hoare 2004; Osborn and

Parker 2003). Innumerable factors can influence the likelihood of a deterrent method being successful

including characteristics of the local elephant population, lay-out of villages and farms, crops grown,

climate, local cultures and traditions, availability of outside assistance, and socio-economics

(examples: Barnes et al. 2015; Chakraborty and Mondal 2013; Gubbi 2012; Webber et al. 2011;

Graham et al. 2010; Campos-Arceiz et al. 2009; Chiyo et al. 2005; Fernando et al. 2005; Sitati et al.

2003). HEC is often seen as an ‘in-trend’ topic for wildlife researchers, attracting international

attention and donor funding, yet whether any meaningful dint in the overall problem has been made

is arguable. This is not for lack of trying or compassion but rather a reflection on the challenging and

changing nature of crop-raiding.

4.2.1 Human-elephant conflict mitigation

Tried and tested methods range from sophisticated government interventions to basic deterrents

making use of common household objects. At the upper end of the scale, methods are often large-

scale, implemented with assistance of government authorities and involve translocating or killing

121

problem individuals or attempting to confine elephants to PA’s using electric fencing (van Eden et al.

2016; Ponnusamy et al. 2016; Fernando et al. 2011; Kioko 2008; Santiapillai et al. 2006; Nelson et al.

2003). When well looked after, electric fencing can be quite effective however it is expensive, requires

frequent maintenance, and does not always account for the natural movement patterns of elephants

or their ecological requirements (Fernando et al. 2011; Perera 2009; Santaipillai 1996). Less success

has been demonstrated by translocations or culling, and serious ethical implications exist for both

(Fernando et al. 2012; Santiapillai 1996).

At the lower end of the scale, and often referred to as ‘traditional’ or ‘farm-based’ methods, are

measures farmers implement themselves. Common techniques include banging on pots or tin drums

and shouting, throwing firecrackers, throwing sticks or clothes doused in petrol and set alight, shining

bright torches, or hanging reflective objects or materials from wire strands (examples: van de Water

and Matteson 2018; Haturusinghe and Weerakoon 2012; Fernando et al. 2011; Hedges and Gunaryadi

2009; Perera 2009; Sitati and Walpole 2006; Thaufeek et al. 2004). Although these methods can be

effective at chasing elephants away, most farmers find elephants habituate over time, with bolder

elephants sometimes ignoring these efforts completely (Fernando et al. 2011; Sitati and Walpole 2006;

Thaufeek et al. 2004; Osborn and Parker 2003). Furthermore, farmers are required to be in close

proximity to the elephants to execute the deterrents. Given that most crop-raiding occurs at night

(Das et al. 2014; von Gerhardt et al. 2014; Graham et al. 2010; Campos-Arceiz et al. 2009;

Wikramanayake et al. 2004; Bandara and Tisdell 2004), and that methods are designed to agitate or

frighten the elephants, the use of ‘traditional’ methods often places farmers in situations fraught with

danger (Madden 2004). Additionally, psychological costs are often incurred due to the stress and

fatigue caused by living in a constant state of high alert and spending long hours guarding crops at

night after working in fields during the day (Barua et al. 2013; Hartter et al. 2011; Ogra 2008; Sukumar

2006; Thirgood et al. 2005; Hill 2004; Hoare 2000).

Out of frustration and desperation, farmers are increasingly resorting to often-illegal methods that

inflict substantial harm or injury to elephants (Haturusinghe and Weerakoon 2012; Santiapiallai et al.

2010; Bandara and Tisdell 2002). Some farmers will lace favoured foods of elephants with poison,

which can have harmful effects on numerous species who may come into contact with it (Haturusinghe

and Weerakoon 2012; Bandara and Tisdell 2002). In Sri Lanka, explosive devices (locally called hakka-

pattas) have been placed in vegetables along the edges of fields, designed to detonate as an elephant

122

bites into the vegetable (Haturusinghe and Weerakoon 2012). Shooting elephants, either directly or

by setting trap-guns, is also quite common, as evidenced by the prevalence of elephants bearing

wounds consistent with gunshot injuries (Haturusinghe and Weerakoon 2012; Bandara and Tisdell

2002). Exactly how frequently such methods occur, or how many people employ them, is difficult to

quantify as farmers very rarely admit to using illegal techniques for fear of repercussions (Bandara and

Tisdell 2002). Some scientists purport that elephants who are exposed to violence or harm at the

hands of humans, are in turn responding more aggressively to humans, thus perpetuating a cycle of

conflict (Fernando et al. 2011; Perera 2009).

In the last two decades, a shift towards community-based eco-deterrents has become more prevalent.

Examples include chili fences, chili-dung briquettes, bio-fences such as Palmyra trees, and buffer zones

of unpalatable crops (examples: van de Water and Matteson et al. 2018; Gross et al. 2017; Changa et

al. 2016; Gross et al. 2016; Karidozo et al. 2015; Baishya et al. 2012; Hedges and Gunaryadi 2009;

Fernando et al. 2008; Graham and Ochieng 2008; Sitati and Walpole 2006; Nelson et al. 2003). Some

have the advantage of exposing approaching elephants to a negative consequence. For example, chili-

based deterrents contain the capsaicin compound which is an irritant to the sensitive eyes, trunk and

mouth of elephants (Changa et al. 2016; Karidozo and Osborn 2015; Hedges and Gunaryadi 2009).

Some may provide an alternative or additional source of income. In Nepal, several species of medicinal

and aromatic plants (e.g. chamomile, coriander, basil, turmeric, lemon grass and citronella) that are

of high market value have been identified as unpalatable to elephants, and their potential as either

strategically planted buffers or alternative crops are being investigated (Gross et al. 2015). Typically,

methods such as these are employed with the assistance of local or international NGO’s however the

premise is that after set-up and any initial training required, farmers will be able to maintain and

expand the deterrents independently (Changa et al. 2016; Karidozo and Osborn 2015; King et al. 2011;

2009). As is the norm with HEC mitigation, each method comes with challenges and advantages, and

success levels are variable (Pozo et al. 2017; Gross et al. 2017; 2016; Baishya et al. 2012; Hedges and

Gunaryadi 2009).

4.2.2 Elephant interactions with bees and beehive fences- Africa

A relatively new community-based eco-deterrent currently generating a lot of interest worldwide is

the beehive fence. Put simply, a beehive fence consists of a series of beehives hung from posts, with

123

each hive spaced about 10 m apart and connected with a single wire. These connected hives are

strategically positioned around an area to be protected from elephants. Should an elephant attempt

to enter the protected area by walking in between beehives, he/she will hit the connecting wire,

causing the hives to swing and disturb the bee colony inside, inciting the bees to exit the hives and

swarm at the elephant (King et al. 2011; 2010; 2009) (Figure 4.1 and 4.2). Beehive fences have the

potential advantage of producing both a direct negative consequence to elephants (bee stings) and

an opportunity for additional income generation through sales of honey and wax products (Scheijen

et al. 2018; King et al. 2017; 2011; 2009; Nair and Jayson 2016).

Figure 4.1. Diagram of a beehive fence © Morgan Tipper

Figure 4.2. A beehive fence © Kylie M. Butler.

124

Understandably, the concept of the thick-skinned elephant being afraid of the comparatively tiny bee

can initially seem absurd. However, in Africa at least, this fear has long been expressed through

folklore and rock art (King 2010). Of more direct relevance are the anecdotes relayed by local farmers

living in proximity to elephants, mostly involving elephants running away from bee swarms after

accidentally disturbing hives (King 2010; King et al. 2009). Scientific evidence supports anecdotal

evidence with field experiments showing that African elephants avoid resting under, or feeding from,

trees containing beehives (Cook et al. 2017; Ngama et al. 2016; Vollrath and Douglas-Hamilton 2002).

Audio playback experiments exposing African elephants to the sounds of disturbed African honeybees

caused elephants to increase vigilance behaviours, to explore the immediate surroundings using

olfaction, to head shake or dust, and to walk quickly or run away from the source of the sound (King

et al. 2007). African elephants also produced a specific rumble vocalisation thought to warn nearby

conspecifics of a threat in the area, and conspecifics responded to audio playback of the rumbles with

similar behavioural responses as when exposed to actual bee sounds (King et al. 2010; 2007).

Capitalising on this apparent fear of bees, Dr. Lucy King developed the beehive fence as a HEC

mitigation tool (King 2010). Encouraging results from a pilot study in Kenya demonstrated a reduction

in crop damage on farms with beehive fences (King et al. 2009) and inspired more trials in other

locations in Kenya (King et al. 2017) and further afield within Africa (Scheijen et al. 2017). At one

location in Kenya, beehive fences showed a capacity to deter up to 80% of crop-raids from small-scale

farms (King et al. 2017). Results from a linear fence in Tanzania aiming to prevent elephants from

entering farmlands bordering PA’s also demonstrated success (Scheijen et al. 2017). That farmers

expressed positive perceptions of participating in trials and additional members requested

involvement was another promising indicator (King et al. 2009). Trials also elicited several successful

honey harvests from which farmers could keep the honey for personal use or sell it at local markets

(King et al. 2017; 2011; 2009).

4.2.3 Elephant interactions with bees and beehive fences- Asia

Positive signs of deterrent success using beehive fences in Africa, has prompted much speculation

regarding its potential applicability as an elephant deterrent in Asia. Despite many obvious similarities

between the HEC situations in Africa and Asia, and particularly crop raiding on small-scale farms, HEC

mitigation tools are not always transferrable. In the example of beehive fences, it is important to

125

recognise that both the Asian elephant and Asian honey bee (Apis cerana indica) species differ from

their African counterparts, and to consider what impact this may have on beehive fence deterrent

success.

While in Africa it is generally accepted that elephants will avoid bees (Cook et al. 2017; Vollrath and

Douglas-Hamilton 2002) less evidence, either anecdotal or scientific, exists in Asia to support the

notion that Asian elephants might also display fear or avoidance in response to the presence of bees.

In northern Thailand, a normally placid captive elephant was observed trumpeting and attempting to

run before being brought under control by her mahout while resting under a tree containing a

disturbed beehive (pers. obs. 2012). However, none of the local elephant owners or mahouts had ever

heard of, or witnessed, elephants being afraid of bees (pers. comm. 2012). In Sri Lanka, informal

conversations with residents at the study site of Dewagiriya Village did not reveal any historical

knowledge of elephants avoiding bees, despite people being familiar with encountering both wild

elephants and wild beehives in the forests (pers. comm. 2014). Still, playback experiments in Sri Lanka

exposing wild Asian elephants to audio of disturbed Asian honey bees, showed elephants responded

with avoidance and exploratory behaviours (King et al. 2017), suggesting there may be potential for

the use of bees as a deterrent.

In Asia, and at our study site in Sri Lanka specifically, two characteristics stand out as potential limiting

factors for the success of beehive fencing. Firstly, the Asian honey bee is comparatively less aggressive

than the African honey bee (Apis mellifera scutellata) (Punchihewa 1994). Also, of the three honey

bee species found in Sri Lanka, Apis cerana is the only one to nest in enclosed cavities (Punchihewa

1994), rendering it the only species possible for use in a beehive fence. However, this honey bee is

sensitive to disturbance, swarming or absconding easily and thus being difficult to attain bountiful

honey harvests from (Koeniger and Koeniger 1994). On the one hand, this behavioural characteristic

could be advantageous for deterrent effect, in that easily disturbed bees might respond aggressively

towards elephants. However frequent absconding could be problematic for overall fence

management and beekeeping.

At present, little is known about the potential of beehive fencing in Asia. To the best of my knowledge,

the only peer-reviewed study is from a small trial in India using 20 beehives to block elephant

pathways into crop fields (Nair and Jayson 2016). Although results were positive, the study is limited

126

by a relatively short nine-month study period and a small sample size with ten of 14 elephant

approaches instigated by the same elephant (Nair and Jayson 2016). The fence design also differed

from the original African beehive fence, as the beehives were placed to block path entrances (Nair and

Jayson 2016) rather than restricting access to crops by surrounding a plot or garden (King et al. 2017;

Scheijen et al. 2017; King et al. 2009; 2007).

Other indicators of beehive fence success in Asia stem from anecdotal newspaper reports (Dutta 2013;

Manoj 2012) and one short three-month trial where a small beehive fence was used to block elephant

access at one corner of a five-acre paddy field (Shri et al. 2013). Whilst encouraging, accurate

evaluations of the deterrent effect of beehive fencing for Asian elephants, require longer-term studies

with systematically collected data. Additionally, cost analyses and identifying farmer willingness to

adopt this method are essential to an overall assessment of the potential role of beehive fencing as

an Asian elephant HEC mitigation tool.

Here we aimed to address this knowledge gap by establishing a trial site of ten beehive fences in a

small village in Sri Lanka heavily impacted by crop-raiding, to assess the potential of beehive fencing

using Asian honey bees as an Asian elephant deterrent. Specifically, we aimed to:

− quantify and compare the frequency of elephant approaches to experimental and control

gardens, and the frequency with which elephants enter experimental and control gardens

− determine the financial costs of establishing and maintaining a beehive fence, and identify

common maintenance issues or challenges

− identify farmer willingness to adopt beehive fences as an elephant deterrent

− discuss overall feasibility of beehive fencing as an Asian HEC mitigation tool based on findings

from this trial

4.3 Methods

4.3.1 Study Site

This study was conducted in the small rural community of Dewagiriya Village (07°33.928' N,

080°56.866' E), located in the dry zone of Central Sri Lanka. Spanning an area of approximately 2 km²

127

and consisting of around 40 households, Dewagiriya Village lies approximately five kilometres south-

east from the nearest boundary of Wasgamuwa National Park, and two kilometres west of the

Mahaweli River (for further detail on the study site refer Chapter 3) (Figure 4.3).

Figure 4.3 Study site location

128

The community of Dewagiriya Village rely primarily on rice farming for subsistence, as a staple of their

diet and/or to generate income. The average household size is 3.21 adults (± SE 0.24) and the average

annual income per household is 131 500 LKR (± SE 18 662, range: 24 000 to 339 000) or 747 USD (± SE

107, range: 136 to 1928). The community is heavily impacted by HEC in the form of crop-raiding, and

all households are negatively affected to varying degrees. Households located to the south of the main

road, which are closest to the water tanks and forest, typically experience the most frequent elephant

visits and crop damage (see Chapter 3). Elephants not only raid crops growing in fields, but they visit

home gardens where most families have banana and coconut trees and plant small fruit and vegetable

plots. Structural damage to houses as elephants attempt to access stored rice inside is a further

consequence. Elephant activity in proximity to homes is a serious safety concern for the Dewagiriya

community.

Beekeeping is not traditionally practised in Dewagiriya Village, however some households place clay

pots in their gardens to attract wild bees. Some community members, mainly older men, have

experience harvesting honey from wild hives in the forest however this is not commonly practiced

anymore. Beekeeping in its modern form using Langstroth’s hives was not practiced by any community

members prior to this study.

4.3.2 Introducing the concept of beehive fencing and participant selection for the beehive fence trial

The concept of using beehive fences to deter elephants from entering and foraging in crop fields or

home gardens was introduced during two community meetings. Prior to this, no community members

were aware of this HEC mitigation strategy nor did anyone have experience or knowledge of elephants

being afraid of bees. All adults in the community were encouraged to attend the meetings and were

notified one week prior by local community leader Mr. B.K.G. Jayasena. Meetings were held in June

and July 2014 and were attended in full by 20 and 17 adults respectively. At both meetings, additional

community members joined for shorter periods in between tending their fields. Dialogue was

translated from English to Sinhalese and vice versa on the spot by local bilingual assistants.

129

At the first meeting local HEC issues, patterns of crop-raiding, and deterrent methods currently used

were discussed. The basic premise of the beehive fence deterrent was explained, and I showed

footage of African elephants responding with avoidance to playback audio of disturbed African honey

bees and photos of beehive fences used in Africa (courtesy of Dr. King, pers. comm). Most attendees

expressed surprise or disbelief that elephants displayed fear or avoidance of bees but thought that if

beehive fences were demonstrating effectiveness in Africa that it would be worth trying in Asia also.

Most of the attendees expressed interest in learning beekeeping skills due to the medicinal properties

of honey and its market value.

At the second meeting, the merits of protecting crop fields versus smaller home garden areas were

discussed. Trials in Africa tend to use the beehive fences to surround crop plots of one to two acres,

however the layout of rice farming in Dewagiriya Village makes it difficult to delineate rice fields into

similar sized plots. Although each family farms between one to three acres, the fields of multiple

families are often contiguous making the overall crop field much larger. Given that elephants also

frequently raid home gardens where various fruits, grains or vegetables grow year-round, and that

home garden boundaries are more clearly defined, the community suggested that home gardens form

the focus of the beehive fence trial. Attendees then identified 18 households in the village which they

deemed to be most affected by crop-raiding and most in need of assistance to defend their gardens

(Table 4.1).

Local field assistant Mr. S. Herath and I visited each listed household individually to assess their

suitability for participation in the beehive fence trial. Having already established that crop-raiding was

a threat to livelihoods and/or safety, suitability was based on the following criteria: a) fruits, grains or

vegetables must be grown in the garden for at least part of each year; b) at least one adult must live

at the home year-round; c) at least one adult must be willing to complete ‘elephant event’ data sheets

and/or communicate verbally details of ‘elephant events’ on at least a bi-monthly basis; d) at least

one adult must be willing to learn beekeeping and actively participate in fence establishment and

maintenance

130

Household name House code Location description

UG Gunapala H01 South of road, west side of village, near water tank

*UG Sobana H02 South of road, west side of village, near water tank

*AG Jaysinghe H06 South of road, west side of village

UG Sirisena H07 South of road, west side of village

GMG Dharmadasa H12 South of road, west side of village

*AG Ariyadasa H20 South of road, mid-village

*PGN Somathilaka H22 South of road but adjoining, mid-village

MG Jothirathne H26 South of road, mid-village

UA Ariyasena H27 South of road, mid-village

BG Sirisena H28 North of road, east side of village

*BKG Jayasena H30 South of road but adjoining, east side of village

UA Rathnasena H31 North of road, east side of village

*WG Dharmadasa H32 South of road, east side of village, near water tank

*UG Guneris H33 North of road, east side of village

*WG Karunadasa H34 South of road but adjoining, east side of village

*BG Sumanawathi H35 South of road, east side of village, near water tank

*LG Ghanawathi H36 South of road, far-east of village, near water tank

UGSK Thilanarathne H37 South of road but adjoining, far-east of village

Table 4.1 List of households experiencing the most negative impacts of crop-raiding in Dewagiriya Village as

determined by 17 community meeting attendees. * next to household name denotes selection for participation in

the beehive fence study as an experimental subject.

In the spirit of a true collaborative community HEC mitigation initiative, and to foster a sense of

ownership and responsibility of individual fences, community preferences were taken into

consideration when selecting the ten households to participate in the experimental beehive fence

trial. Although an ideal experimental design would see beehive fences distributed evenly throughout

the village, the community was insistent that those most affected by crop-raiding and in need of the

most help, should be the experimental participants. Each experimental home garden was paired with

a control home garden of approximately the same size and growing similar fruits, grains or vegetables.

131

Where possible, a neighbouring farm was selected however where this was not practicable a non-

neighbouring farm of the same north or south section and of similar proximity to forest and water

tanks was chosen (Figure 4.4). As for experimental participants, selection criteria b) and c) also applied

to control households however d) was not relevant.

At all stages of experimental participant selection, I clarified that it was unknown whether beehive

fences using Asian honey bees would successfully deter Asian elephants and that this was an

experimental trial to evaluate whether beehive fences had potential as an effective Asian elephant

deterrent.

Figure 4.4. Map of Dewagiria Village showing experimental and control households. Experimental households

are denoted with a black diamond plus their identification code. Control households are denoted with a black

circle plus their identification code. Experimental and control households can be identified using the

corresponding numbers © Morgan Tipper.

132

4.3.3 Establishment of beehive fences

The beehive fence design used in Dewagiriya Village followed the basic protocol outlined in the

‘Beehive Fence Construction Manual’ (King 2014) and trialled in Kenya (King et al. 2017; 2011; 2009),

with some alterations made to better suit the Sri Lankan environment and make use of locally available

materials. The beehives used were ten-frame Langstroth hives with a single honey super on top (Figure

4.5). Langstroth hives were favoured over traditional log or clay pot hives as they were easier to hang

and more ‘user friendly’ in terms of monitoring hive health and harvesting honey. Also, they were

readily available for purchase from Sri Lankan beekeepers and relatively inexpensive (14.50 USD/1 900

LKR). Initially, hive parts were held together using coconut rope however during strong wind or rain

hive compartments tended to slip. To better secure hive parts, we sourced ‘z’ clips from Australia

which were far more effective, particularly during the monsoon (figure 4.6).

Hooks were attached to the sides of beehives and fencing wire was used to hang the hives from strong

wooden posts at a height of approximately 1.5 m. Each post was between 2.5 and 2.8 m length and

was treated with old engine oil, as an alternative to insecticides, for protection against termites prior

to fence building (figure 4.7).

LEFT: Figure 4.5 Langstroth beehive used in the beehive fence and secured with coconut rope ©

Kylie M. Butler. RIGHT: Figure 4.6. Langstroth beehive with the ‘z’ clips used to secure hive parts ©

Kylie M. Butler

133

Posts were embedded in 0.75 m holes with stones and soil packed around the base. Each beehive was

hung between two posts spaced approximately 3 m apart. Neighbouring beehives were approximately

8 m distance away and were connected to one another using single strands of strong, taut fencing

wire. Above each beehive, a shade roof was hung (figure 4.8).

Figure 4.7. Painting fence posts with old engine oil to protect from termites –

researcher Kylie M. Butler and field assistant Supun Herath © Morgan Tipper.

Figure 4.8. A beehive fence at the Jayasena household. © Kylie M. Butler

134

Over the course of the trial four types of shade roof were used: a) banana leaf on a stick frame; b)

dried grass on a stick frame; c) thin plywood with waterproof paint; and d) thick plywood with

waterproof paint (figure 4.9). Finally, although most beehive fences in Africa alternate ‘dummy

beehives’ (wooden boards cut to the same dimensions and painted the same colour as beehives) and

real beehives, it was decided that only real beehives would be used to help the beehive fences achieve

the strongest deterrent effect possible.

A total of ten beehive fences were built between October 2014 and September 2015. With the

assistance of Mr. Herath, I worked closely with each household to design individual fence layouts

taking into consideration: a) the direction elephants typically approached; b) any natural barriers such

as thick or tall vegetation; c) the boundaries of their home garden; and d) location of seasonal

vegetable or crop plots that might make it difficult to access the fence for maintenance or beekeeping.

As a result, the exact shape and size of each fence varied somewhat (Figure 4.10), with the smallest

Figure 4.9. Beehive fence shade roofs. Top Right: Mr. Somathilaka standing by his beehive fence with a traditional

woven banana leaf shade roof. Top Left: A ‘rice’ grass shade roof on the Guneris family’s beehive fence. Bottom

Right: A thin plywood shade roof warping on the Ariyadasa family’s fence. Bottom Left: A thick plywood shade roof

on the Somathilaka family’s fence. © Kylie M. Butler

135

fence using ten beehives and the largest two fences using 18 beehives each. Some fences formed a

full square around the home garden, and others a ‘C’ shape utilising natural vegetation barriers along

certain boundaries. On average, beehive fences took approximately eight hours to build, usually over

two mornings. A team of at least five people, including one or more adults from the given household,

worked together on all aspects of fence construction from digging the holes for the posts to making

the shade roofs.

Figure 4.10. Layout of the ten beehive fences.

136

4.3.4 Data collection: elephant events

Data collection was conducted between January 01st 2015 to December 31st 2017, with precise dates

varying depending on establishment dates of individual beehive fences. Data collection for

experimental participants whose beehive fences were complete in 2014 began on January 1st 2015.

Two households elected to discontinue participation in the trial partway through and data collection

ceased on the date their beehive fence was dismantled. Data collection for control home gardens

followed the same dates as for their paired experimental home garden (Table 4.2).

.

Household Name

Code Fence established

Data collection period Data collection duration

Somathilaka E1 2014-10-28 2015-01-01 to 2017-12-31

36 m

Sobana E2 2014-11-05 2015-01-01 to 2017-12-31

36 m

Ariyadasa E3 2014-11-14 2015-01-01 to 2017-07-15

30 m, 2 w, 1 d

Jaysinghe E4 2014-11-14 2015-01-01 to 2017-12-31

36 m

Ghanawathi E5 2015-03-03 2015-03-03 to 2017-04-20

25 m, 2 w, 4 d

Guneris E6 2015-04-02 2015-04-02 to 2017-12-31

32 m, 4 w, 2 d

Sumanawathi E7 2015-05-23 2015-05-23 to 2017-12-31

31 m, 1 w, 2 d

Dharmadasa E8 2015-06-06 2015-06-06 to 2017-12-31

30 m, 3 w, 5 d

Jayasena E9 2015-09-01 2015-09-01 to 2017-12-31

28 m

Karunadasa E10 2015-09-07 2015-09-07 to 2017-12-31

27 m, 3 w, 4 d

Table 4.2. Fence establishment and data collection periods for each experimental and control household

137

Following the same protocol as previously detailed in ‘Chapter 3’, data collection utilised resident

reports of ‘elephant events’ collected during at least bi-monthly home visits to each experimental and

control household. However, here an ‘elephant event’ was defined as any sighting of one or more

elephants within 50 m of the home garden, as well as any successful crop-raids or property damage

that occurred within the home garden boundary. The same data sheets were used and again,

participants preferred to relay information verbally during visits rather than using the provided data

sheets, and this preference was accommodated.

4.3.5 Fence maintenance

At least one adult per experimental household was trained in basic fence maintenance. This involved

a bi-monthly walk around the beehive fence perimeter to ensure the following: a) that wires

connecting neighbouring hives were attached; b) posts were upright and strong; c) hives were hanging

approximately 1.5. m from the ground and not askew, and that shade roofs were not touching the

hives; d) empty hives were free of pests or debris; e) shade roofs were not damaged or falling apart;

and f) the area underneath beehives was clear of tall vegetation (Figure 4.11). It was recommended

that small tasks (e.g. reattaching connecting wires and straightening hives) were completed during the

check, while larger tasks (e.g. replacing fence posts or making new shade roofs) were scheduled for

completion as soon as possible.

However, although participants were requested to complete fence checks regularly, with support

available for any maintenance requiring new materials, most households failed to do this. As accurate

assessment of beehive fence deterrence required fences to be in good condition, our field team

performed fence checks at least monthly, asking farmers to assist when they were home. All

maintenance issues and tasks completed were documented on data sheets detailing the date,

description of maintenance issue, and whether it had been rectified.

138

Figure 4.11. Fence maintenance issues. Top Right: Researcher Kylie M. Butler, field assistant Supun Herath

and Mr. Sumanawathi raising the shade roofs. Top Left: Mr. Somathilaka straightens a crooked beehive. Centre:

Mr. Guneris repaints his post with engine oil to protect against termites. Bottom Right: A mouse escapes from a

beehive. Bottom Left: Chew damage from mice in an occupied beehive. All photos © Kylie M. Butler except Top

Right © Morgan Tipper

139

4.3.6 Beekeeping

At least one adult from each experimental household was trained in basic beekeeping skills including:

safe conduct around bees, colony inspection to assess strength and health, feeding sugar water,

attracting wild bees, transferring and splitting colonies, and honey extraction. Initial training

throughout 2015 consisted of home visits with a local beekeeping expert. In 2016 and 2017, follow-

up workshops and home visits were conducted by both international and local beekeepers (Table 4.3).

On-call support from our field team was available throughout the study period. Each experimental

household was provided with protective equipment including 2 x beekeeping face veils and hats,

beekeeping gloves and a smoker. Multiple beekeeping jackets were distributed throughout the village

for farmers to share when working with more active or aggressive hives.

Participants were encouraged to actively assist with occupied hive inspections at least quarterly.

Records of colony introductions or abscondments, hive checks and honey harvests were documented

using simple data sheets.

Dates Beekeeper Training type Skills covered

2015-03-16 and 2015-03-17

Dr. R.W.K. Punchihewa

Home visits Hive inspection, identifying comb parts, extracting pests, feeding sugar

water

2015-05-11, 2015-05-12 and

2015-05-25

Dr. R.W.K. Punchihewa

Home visits Introducing colonies, smoking bees and hive inspections

2015-11-24 and 2015-11-25

Dr. R.W.K Punchihewa

Home visits Checking hive health, feeding sugar water

2016-10-03 to 2016-10-07

Dr. D. Gormley O-Brien and

Miss E. Collins

Home visits and group workshop

Hive inspections and identifying issues (pests, weak colonies), feeding bees,

working with aggressive colonies

2017-03-01 and 2017-03-02

Mr. F. Ryde Home visits and group workshop

Attracting wild swarms, splitting colonies, transferring colonies from

clay pots, basic honey harvesting

2017-05-31 and 2017-06-01

Mr. F. Ryde Home visits and group workshop

Splitting and transferring colonies, honey harvesting

Table 4.3 Beekeeping training and workshop record

140

4.3.7 Cost analysis

In order to determine the overall costs of building and maintaining a beehive fence, and to assess its

affordability for small-scale rural farmers in Sri Lanka, all expenses incurred were recorded. This

included all fencing materials, basic tools, basic beekeeping protective equipment, colonies and colony

delivery, and replacement materials required for maintenance. A minimum cost was then calculated

based on initial set-up requirements for a 50% hive occupation rate.

4.3.8 Farmer acceptance and willingness to participate in using beehive fencing

Farmer acceptance of, and willingness to participate in, using beehive fencing to deter elephants was

assessed through one-on-one social research interviews, informal conversations, and researcher

observations. One-on-one interviews were conducted at the participants home with a local translator.

These involved a series of statements ranked by Likert scale, relating to the perceived effectiveness

of, and interest in continuing to use, beehive fencing. Informal conversations, translated by field

assistant Mr. Herath, occurred while visiting households for data collection or working together on

fence maintenance or beekeeping tasks. These conversations enabled a free-flowing exchange of

ideas and opinions in a relaxed setting. Finally, researcher observations were incorporated into the

overall assessment, as verbally expressed enthusiasm for beehive fencing by participants was not

always in-line with on-ground action.

It must be acknowledged these modes of data collection are largely subjective and reliant on

anecdotal information. However, given the aim of implementing a genuine participatory community-

based deterrent, information gained through frequent interactions with participants in a relaxed

setting has particular relevance. Still it is acknowledged that such qualitative data should be

interpreted with some caution.


To identify whether a difference in frequency of elephant events occurred between experimental

(beehive fences present) and control (beehive fences absent) farms, a paired-sample t-test was used.

To assess whether elephants were less likely to enter a family’s home garden area when a beehive

141

fence was employed, I used Pearson’s chi-square test of independence (χ²) with Yate’s continuity

correction.

To assess whether the frequency with which elephants breached a fence perimeter was influenced by

the amount of hives overall, or specifically the number of occupied hives, I used a generalised linear

mixed model (GLMM), assuming a poisson distribution. The number of hives present, and the number

of occupied hives, were included as predictor variables, with the farm location fitted as a random

intercept to allow for the repeated-measure comparison of fences containing different levels of hive

occupations at different times. Pearson correlations were used to assess collinearity between

predictors before inclusion in the model. Model fit was determined using the marginal and conditional

r-squared (Nakagawa and Schielzeth 2013).


Generalised Linear Mixed Models (GLMM’s) were constructed using the package ‘lme4’ (Bates et al.

2014), with model predictions generated using the ‘effects’ package (Fox 2003). All visual

representations were constructed using ‘ggplot2’ (Wickham 2016). Results were considered

significant when p <0.05.

4.4 Results

4.4.1 Elephant events and deterrent effect of beehive fences

A total of 176 elephant events were reported by experimental (64%, n = 122) and control (36%, n =

64) households, with elephant events occurring significantly more frequently at experimental home

gardens (M = 11.2, ± SE = 1.8, n = 122) compared to control home gardens (M = 6.4, ± SE = 0.90, t(9) =

2.71, p = 0.02) (Figure 4.12).

The likelihood of elephants physically entering home gardens also differed. Here, elephants entered

control home gardens on 83% (n = 64) of approaches compared to only 43% (n = 122) for experimental

home gardens. This shows a significant increase in the likelihood of elephants breaching the home

garden boundary of unfenced control gardens (M = 85.66%, ± SE 4.21%) compared to the beehive

fenced perimeter of experimental farms ((M = 37.55%, ± SE 6.69%) (χ² (1, N = 176) = 24.98, p <0.01)

(figure 4.13).

142

Figure 4.12 Number of elephant events: Beehive fenced (experimental) versus non-beehive fenced

(control) home gardens

143

Figure 4.13: Frequency of elephants entering gardens during an elephant event after approaching a beehive

fence (experimental) or non-beehive fenced (control) home garden.

144

The number of physical beehives used on a fence was not found to influence whether elephants

breached the fence perimeter. However, the presence of occupied beehives did and fences containing

higher numbers of occupied beehives were significantly less likely to experience breakthroughs (Table

4.4, Figure 4.14).

Predictor variable Coefficient SE z p-value R² (marginal) R² (conditional)

(Intercept) 0.05 0.78 0.07 0.95 0.22 0.46

# hives present 0.10 0.05 1.83 0.07

# occupied hives -0.12 0.05 -2.26 0.02

Table 4.4. GLMM summary: effect of total hive numbers and number of occupied hives on fence break-throughs

by elephants. Parameter coefficients, standard errors (SE), z-values and p-values of the fixed terms (marginal)

and fixed and random terms (conditional). Coefficients for which the 95% confidence interval does not include

zero are shown in bold.

145

4.3.1 Fence maintenance

The three most frequently occurring maintenance issues were: pests inhabiting beehives; shade roofs

sagging and touching beehive roofs; and shade roofs becoming damaged and needing replacement.

Based on presence/absence counts per beehive, the most common pests found were mice, ants and

spiders (table 4.5).

Figure 4.14 Relationship between elephant breakthroughs and number of occupied

beehives

146

Year Ants Mice Lizards Frogs Spiders Wasps Snakes Other

2015 55 52 19 11 25 0 2 5

2016 55 86 15 3 21 5 0 15

2017 31 58 8 3 11 6 6 3

Total 141 192 42 14 57 11 8 23

Table 4.5. Presence/absence counts of different pests found in beehives. Bold text denotes the three most

frequently occurring pests per year and overall. The ‘Other’ category includes cockroaches, beetles, palm

squirrels and a bird. Whilst the number of beehives present throughout the study varied according to fence

establishment dates and due to two fences being dismantled in 2017, all hives present were checked monthly for

each of the listed pests. Thus, the table gives an indication of which pests most frequently occurred each year of

the study, as well as overall.

The shade roofs required raising and straightening at least monthly. This involved tightening the wires

around the cross bars, took less than a minute per hive and could be done quite easily by one person.

Shade roofs were frequently damaged by wind or rain and two or more shade roofs per fence required

replacing on 60 occasions. In September 2015 and March 2016, all original grass or banana leaf shade

roofs needed replacing. By August 2016, all grass roofs had been replaced with thin plywood however

these warped during the rainy season. Thick plywood roofs were installed instead in February 2017,

which were more durable but still vulnerable to cracking in the heat.

Other maintenance issues included: farmers leaving connecting wires unattached, overgrown

vegetation underneath hives, beehives hanging askew, damaged beehives or hive parts, and damaged

or weak posts. Fire-damage occurred accidentally at four fences when farmers burnt the land adjacent

to the fence in preparation for planting. All fence posts were replaced at least once during the three-

year period, as a result of termite damage or general weathering.

All households experienced the same types of fence maintenance issues, however those who checked

their fences regularly and rectified issues promptly spent less time on maintenance overall.

Quantifying precise durations of time spent on maintenance was difficult given the different sizes of

fences, capacities of households, and specific maintenance tasks required. However, based on our

experiences assisting with maintenance and information from participants, a fence check including

rectification of basic maintenance issues took between one to two and a half hours with two people,

147

and was required bi-monthly. Larger maintenance issues, such as replacing shade roofs or posts,

needed to be addressed approximately every three months, and took between four to eight hours

with a team of four people.

4.4.3 Beekeeping and honey harvests

The total number of beehives across all ten fences was 142 and of these, 76 (53.5%) hives were

occupied at least once. Overall, 100 colonies occupied beehives. Of these, 23 occupations occurred

naturally when wild honey bees entered the fence hives, and a further three colonies were transferred

into fence hives after being found in clay pots. Despite efforts to attract wild colonies by placing clay

pots and traditional log hives in gardens, wild honeybee occupation rates were low. Thus, most hive

occupations resulted from purchased colonies which were sourced from Kandy or Colombo then

transported to Dewagiriya Village where they were transferred into fence hives. Overall, 74 colonies

were introduced this way.

Of the 100 colonies, 63 absconded before the end of the trial. Excluding failed introductions, where

the colony absconded within seven days of hive transfer, the mean duration of occupation by colonies

was 201 days ((SE ± 26.5). The number of hives occupied at least once, the number and mode of colony

introductions, and the number of abscondments varied per fence (Table 4.6).

All households experienced periods where no active hives were present on their fences, particularly

following fence establishment until the first wild occupation occurred or purchased colonies could be

sourced. Additionally, four fences experienced periods where the number of absconded colonies

exceeded introductions and active hive presence reverted back to 0%. Overall, the mean ‘active hive’

rate per fence was 17.8%. Active hive rates were somewhat biased by an uneven distribution of

purchased colonies. Given the difficulties associated with sourcing and transporting healthy colonies,

and the expense of purchase, more colonies were allocated to beehive fences where participants

demonstrated active involvement in fence maintenance and beekeeping, rather than equally dividing

colonies amongst participants (Table 4.7).

148

Fence code

# hives

# hives occupied ≥ 1

x

# introduced colonies

# wild colonies

# abscondments

Mean occupation duration (days)

E1 12 6 8 4 10 242

E2 17 14 13 5 9 232

E3 17 2 2 0 2 61

E4 12 8 7 5 9 80

E5 10 2 2 0 1 375

E6 11 8 7 1 3 286

E7 18 10 9 4 7 216

E8 15 10 10 2 12 280

E9 12 6 6 3 4 133

E10 18 10 10 2 6 127

Totals 76 74 26 63 M = 203

Table 4.6. Asian honey bee colony introductions and abscondments

Fence code

Build date Days to 1st occupation

Occupation Mode

Fewest active hives

(%)

Most active hives (%)

Mean active hives (%)

E1 2014-10-28 206 Introduced 0 25 20.1

E2 2014-11-05 125 Wild 5.9 70.6 31.7

E3* 2014-11-14 194 Introduced 0 11.8 0.8

E4 2014-11-14 187 Wild 0 41.7 11.8

E5* 2015-03-03 87 Introduced 10 20 13.6

E6 2015-04-02 59 Introduced 18.2 72.7 24

E7 2015-05-23 6 Introduced 11.1 38.9 20.3

E8 2015-06-13 3 Introduced 6.7 60 20.2

E9 2015-09-01 273 Introduced 0 41.7 16.4

E10 2015-09-07 267 Introduced 5.6 50 19.1

Table 4.7. Number of active hives per fence

149

Honey production was low and only six households harvested any honey throughout the trial, totalling

approximately seven litres. The most honey harvested from any individual beehive fence was two

litres and no honey at all was harvested from four fences (table 4.8).

Fence code Honey harvested (L) Month/year

E1 0.5 Mar 2017

E2 0.5 Feb 2017

E3 0

E4 0

E5 0.5 Mar 2017

E6 2 Mar 2017, Sep 2017, (1 L each)

E7 1.0 Mar 2016

E8 0.5 Mar 2017

E9 0

E10 2 Mar 2017

Total 7

Table 4.8 Honey harvesting records per household

Common challenges impacting beekeeping across all experimental fences included: lack of adequate

shade, insufficient food and water resources for bees, pests entering hives (including ants and mice);

and wax moth invasions. Although it is difficult to quantify the precise effects of the aforementioned

challenges, it is reasonable to assume that each contributed to the low rate of wild occupations, the

high instance of colony abscondments, and poor honey production.

4.4.5 Cost of beehive fences

The initial set-up cost of each fence varied somewhat depending on precise fence layout. Efforts were

made to minimise costs by using local materials and labour.

Estimates for the minimum include: beehives (15 USD, 2 626 LKR each), posts (3 USD, 525 LKR each),

cross bars (0.50 USD, 87 LKR each), shade roofs (3.50 USD, 612 LKR each), strong wire for hanging and

connecting beehives (3.50 USD, 612 LKR, p/kg), thin wire for hanging shade roofs (1.35 USD, 236 LKR,

p/kg), and ‘z’ clips to keep moveable hive parts secure (1 USD, 175 LRK, each) (Table 4.9).

150

Item costs (USD) p/BH

Beehives 15.00

Posts 6.00

Cross bars 1.00

Shades 2.75

Fence wire 1.75

Thin wire 0.70

‘z’ clips 4.00

Totals

Fence code

# beehives

Cost p/item p/fence

E1 12 180 72 12 33 21 8.40 48 374.40

E2 17 255 102 17 46.75 29.75 11.90 68 530.40

E3 17 255 102 17 46.75 29.75 11.90 68 530.40

E4 12 180 72 12 33 21 8.40 48 374.40

E5 10 150 60 10 27.5 17.50 7.00 40 312.00

E6 11 165 66 11 30.25 19.25 7.70 44 343.20

E7 18 270 108 18 49.50 31.50 12.60 72 561.60

E8 15 225 90 15 41.25 26.25 10.50 60 468.00

E9 12 180 72 12 33 21 8.40 48 374.40

E10 18 270 108 18 49.50 31.50 12.60 72 561.60

Totals 2130 852 142 390.50 248.5 99.40 568 4430.4

Table 4.9. Cost of beehive fence materials and equipment per fence

Most of the tools required for building and maintaining the beehives fences are common garden tools

which farmers already own however at a minimum each household was supplied with: two hammers

(6.00 USD, 1 050 LKR), three kilograms of nails (4.50 USD, 786 LKR), one digger (9.50 USD, 1 663 LKR),

a machete (4.00 USD, 700 LKR), pliers (5.00 USD, 874 LKR), two paint brushes (4.00 USD, 698 LKR), two

litres waterproof paint (6.00 USD, 1 047 LKR) and two litres old engine oil (7.50 USD, 1 308 LKR). Thus,

the minimum total spent on tools and equipment per beehive fence was 46.50 USD (8 137 LKR).

Basic personal protective equipment for beekeeping was also essential and the minimum supplied to

each household was: two face veils and hats (33.00 USD, 5 778 LKR), two pairs of beekeeping gloves

(31.00 USD, 5 424 LKR), and a smoker (70 USD, 12 248 LKR), totalling 134 USD (23 450 LKR) per beehive

fence.

Finally, given the low rate of naturally occurring hive occupations from wild colonies, the purchase of

colonies was necessary. Colony prices varied between 25 USD (4 371 LKR) and 30 USD (5 246 LKR)

depending on market availability and the supplier. Transport of colonies cost approximately 160 USD

(27 979 LKR) for a delivery of 20 colonies. Bees also needed to be fed with sugar water after first being

introduced and during periods of low rain. The cost of sugar was approximately 0.75 USD (131 LKR)

p/kg with each colony requiring approximately ½ kg of sugar per month (Table 4.10).

151

Colonies

Transport

Sugar

Cost estimate USD 27.50 ea 160 p/20 colonies

0.75 p/kg

Fence code # colonies for 50% occupation

Cost per item per fence

Totals

E1 6 165 48 18 231 E2 9 247.50 72 27 346.5 E3 9 247.50 72 27 346.5 E4 6 165 48 18 231 E5 5 137.50 40 15 192.5 E6 6 165 48 18 231 E7 9 247.50 72 27 346.5 E8 8 220 64 24 308 E9 6 165 48 18 231 E10 9 247.50 72 27 346.5

Totals 2007.5 584 219 2810.5 Table 4.10 Cost estimates for colony establishment and feeding per fence

Fence code BHF materials Tools/equipment PPE Colonies (incl. transport and sugar)

Totals (USD)

E1 374.40 46.50 134.00 231.00 785.90 E2 530.40 46.50 134.00 346.50 1057.40 E3 530.40 46.50 134.00 346.50 1057.40 E4 374.40 46.50 134.00 231.00 785.90 E5 312.00 46.50 134.00 192.50 685.00 E6 343.20 46.50 134.00 231.00 754.70 E7 561.50 46.50 134.00 346.50 1088.50 E8 468.00 46.50 134.00 308.00 956.50 E9 374.40 46.50 134.00 231.00 785.90 E10 561.60 46.50 134.00 346.50 1088.50

Totals (USD) 4430.40 465.00 1340.00 2810.50 9045.90 Table 4.11. Cost estimate for fencing materials, tools/equipment and colonies overall (per fence)

Accounting for all the expenses detailed above, the cost of establishing the physical beehive fence,

introducing colonies to reach a 50% hive occupation target, supplying households with protective

beekeeping equipment and feeding sugar water to bees ranged between 685.00 USD (119 875 LKR)

and 1088.50 USD (190 488 LKR), with a mean cost of 904.57 (±SE 50.58) (158 299 LKR, ±SE 8 851)

(Table 4.11). Replacement of posts and plywood shade roofs when necessary (estimated 18-month

intervals if well looked after) are additional on-going expenses that need to be considered.

152

Additionally, if bee colonies abscond and need to be re-introduced from suppliers, this adds

substantially to ongoing costs.

4.4.6 Farmer acceptance and willingness to participate

Enthusiasm for trialling beehive fences as a HEC mitigation tool was expressed during initial

community meetings and further demonstrated by over half of the households in Dewagiriya Village

requesting to participate. Still, people were unsure whether Asian honey bees would actually have a

deterrent effect on elephants, and at least part of people’s enthusiasm in the trial could be attributed

to the prospect of harvesting honey for personal use or income generation.

One-on-one interviews conducted in October 2017 with all ten experimental households, revealed

that only one found their beehive fence to be ‘very’ effective, while a further five households stated

the beehive fence was ‘somewhat’ effective. Five households believed elephants were deterred by

the physical structure of the fence as the connecting wires resembled those on an electric fence,

rather than being deterred by the presence of bees. Only two households felt that Asian elephants

were afraid of Asian honey bees.

Participants from all households said they had the theoretical knowledge to look after their fences but

were limited by physical skills or resources, resulting in two households withdrawing from the trial in

2017. Of the remaining eight households, only four said they ‘definitely’ planned to continue using

their beehive fences. The other four felt that without on-going in-field support they did not have

sufficient beekeeping skills. No household had the financial means to replace colonies or broken fence

parts such as posts or hives. However, all eight remaining households stated they wanted to continue

beekeeping for honey production, despite low honey yields during the trial.

Researcher observations concurred with expressed challenges in the capacity to independently look

after fences but also found motivation to be an issue. By the end of 2017, only three households

consistently kept up with basic maintenance requirements and beekeeping. Of the remaining five

households, two were limited in their ability to care for their fences due to being physically incapable

or lacking time. The remaining three households appeared to simply not put in the minimum effort

required. Overall, fluctuations in motivation were evident throughout the trial. Peaks were observed

153

after initial fence establishment and following colony introductions. Troughs were observed in the

period between building the fences and introducing the first colonies, and when hive occupations

were low. Farmers were much more active when working alongside our team, or when consistently

reminded what to do.

4.5 Discussion

Given that crop-raiding has been a recognised HEC issue for decades (Fernando et al. 2005; de Silva

1998; Santiapillai 1996), it is no surprise that a variety of deterrent techniques have been trialled and

that many of them have demonstrated success in deterring elephants (Baishya et al. 2012; Gunaryadi

et al. 2017; Davies et al. 2011; Thapa 2010; Perera 2009). It’s then fair to pose the question: why does

HEC, and specifically crop-raiding, remain such a major issue negatively impacting both elephant and

human communities?

One reason could be that most studies either focus specifically on, or publish information primarily

about, the success of the deterrent effect by quantifying elephant visits or damage incurred (Nair and

Jayson 2016; Baishya et al. 2012; Gross et al. 2017; Gross et al. 2016; Santiapillai 1996). Of course, if

the goal is to prevent elephants from entering a certain location or damaging certain crops, the

capacity of the given deterrent to do this, is a major factor in its success, and an important first step.

However, numerous factors including labour intensiveness, cost-effectiveness, and stakeholder

acceptance, also influence the success of a deterrent in a practical context. More and more studies

are incorporating an increasingly holistic approach into the evaluation of HEC mitigation techniques

(Gunaryadi et al. 2017; King et al. 2017; Changa et al. 2016; Dickman 2010; Graham and Ochieng 2008;

Sitati and Walpole 2006). It is hoped that this will be benefit management planning by facilitating

identification of appropriate deterrent techniques for various HEC locations, that are effective not

only in theory but in practice.

We incorporated a holistic approach in the overall evaluation of our three-year beehive fence trial in

Dewagiriya. The capacity of the beehive fences to deter elephants from people’s home gardens,

maintenance and beekeeping challenges, cost-effectiveness, and farmer motivation were all

important factors in determining whether beehive fencing had potential to help the local community

alleviate their problems with elephants. Taking these factors together, we identified some positive

154

indicators of success but also highlighted several challenges that need to be considered carefully if this

technique is to be recommended as an Asian HEC mitigation tool.

If deterrent effect is the first indicator of success, the beehive fence trial in Dewagiriya ticked this

initial box. Results showed evidence that the beehive fence had the capacity to deter Asian elephants

from areas within the fence perimeters. Significantly less elephant incursions occurred in home

gardens surrounded by beehive fences compared to those without. The deterrent effect also appeared

to strengthen with increased hive occupations, which is a promising indicator that elephants were

deterred by the presence of bees rather than simply the physical fence structure.

Still, elephants entered home gardens of beehive fenced properties on 43% of approaches. Although

this is a vast improvement on the 83% of elephant entries that occurred at non-fenced gardens, the

question remains whether this is a sufficient reduction to warrant wider implementation of beehive

fences throughout the village. Elephants often cause considerable damage during even a single crop-

raiding event to crop or vegetables plots, fruit trees, or even houses. A much higher deterrent rate

would be required before community members could relax their vigilance and be confident the

beehive fences could keep them and their gardens safe.

No comparable beehive fence trials exist in Asia, from which to judge whether a 57% deterrent rate

indicates good, bad or average performance. Our methods were based on research trials located in

Kenya, with African honey bees and African elephants. Here, a deterrent success rate of approximately

80% was observed, and this is a clear positive benefit to crop protection (King et al. 2017). Ideally, we

would have liked to have seen a similar success rate in Dewagiriya Village however biological

differences in the aggression levels of the two bee species was always a concern, with the Asian honey

bees comparatively less aggressive (Punchihewa 1994). Based on responses to audio playback

experiments, Asian elephants do elicit avoidance and exploratory responses to the perceived presence

of bees however with less intensity than African elephants (King et al. 2018). It is therefore not

surprising that the deterrent effect observed in beehive fence trials across the two continents

mirrored this pattern. Expanding on the anecdotal newspaper reports (Dutta 2013; Manoj 2012) and

short-term trial in India (Nair and Jayson 2016) by establishing further beehive fence trials following a

similar protocol to King et al. (2011; 2017) and using a multi-factor evaluation process would provide

155

a basis for comparison and help make an informed decision about the worthiness of pursuing beehive

fences using Asian honey bees as a deterrent technique.

The lower levels of aggression weren’t the only challenge presented by bees in Dewagiriya Village.

Natural colonies were difficult to attract into fence hives. On advice from local beekeeping experts,

traditional log and clay-pot hives were strategically placed near homes to attract wild swarms, and

bees wax smeared on hive entrances. Still, only 26 of 100 hive occupations occurred naturally,

rendering the purchase of colonies from Kandy or Colombo necessary, and adding additional expense

to the project budget. Colony health was often weak, indicating that local food sources were

insufficient, even during the rain seasons. We had only ten fences in the village each with between

ten and 18 beehives, yet the mean hive occupation rate per fence was only 17.8% which was far below

our target of 50% occupation rates per fence. This begs the question as to what the saturation point

for supporting bee colonies in Dewagiriya Village would be, and had we already exceeded it? Further

research investigating what level of hive occupation is required to consistently deter elephants, and

how the local environment could be improved to better support bee health, would assist in

determining how feasible it would be to expand beehive fences in Dewagiriya Village.

Whether a deterrent is financially viable for a community is another important consideration. The

beehive fence is described as a low-cost, low-maintenance community based-deterrent (King et al.

2011) and compared to large-scale techniques such as electric fencing, expenses are relatively low.

Still, HEC disproportionately affects poorer communities and it is vital that costs are considered in the

context of the economic status of the community in which it is being proposed. In Dewagiriya Village,

the mean cost of building a beehive fence and introducing a 50% rate of hive occupations exceeded

the mean annual household income (904 USD versus 747 USD). Other studies suggest that economic

benefits from honey production and other bee products could help off-set costs (King et al. 2017).

However, in Dewagiriya Village honey production was low and no economic benefits were generated

from this. The most honey any single household extracted over the three-year period was 2 kg’s and

while farmers enjoyed the small quantities of honey, it was nowhere near a level required to

contribute financially. Realistically, outside funding to support the establishment of beehive fences

would be essential for most communities requiring one.

156

The last factor incorporated into our evaluation was farmer motivation and willingness to participate.

The largely qualitative and subjective nature of assessing people’s motivations makes it difficult to

assess, and even more difficult to compare across studies. Still, it is too important to discount. For

community-based mitigation, regardless of how strong the deterrent effect of a particular technique

is, if farmers have no interest in actively using the technique, it will undoubtedly fail.

In Dewagiriya Village, the motivation of participants ebbed and flowed. Farmers verbally expressed

enthusiasm for the project however this did not always translate into active participation on-ground.

During important milestones, such as establishing the first fences and introducing colonies, members

of all households were actively involved. However, during lulls in hive occupations or when people

were busy planting or harvesting their fields, beehive fences were often neglected. To ensure fences

were well-maintained so we could properly assess deterrent effect, our field team often did

maintenance work alone. This perhaps led to expectations that households did not have to take

responsibility for their own fences. Farmers who were enthusiastic about beekeeping took the best

care of their fences. Overall, only two households were capable of looking after their fences

independently, and two households withdrew from the trial in 2017. When asked if they wanted to

continue using their fences after the three-year trial period, only four households said yes. Most felt

they could not manage their bees without assistance, and that costs of even small repairs would

unaffordable.

When assessing the feasibility of using chili-based deterrents to combat crop-raiding in Indonesia,

Hedges and Gunaryadi (2009) found that participation rates were higher if farmers had confidence in

the effectiveness of the proposed deterrent technique. In Dewagiriya Village, people seemed hopeful

that the beehive fences would help deter elephants but also expressed scepticism about whether it

would work. All participating households said they were motivated by the potential benefits of

harvesting honey. King et al. (2017) found that Kenyan farmers tended to fence maintenance issues

promptly as they recognised the tangible economic benefits from honey yields. Although not explicitly

stated, it is likely low honey production negatively impacted motivation to maintain fences.

Another potential issue with farmer motivation is that engendering a sense of ownership and

responsibility for HEC mitigation with a community can require a major shift in thinking, especially if

they are used to relying on outside assistance (Osborn and Parker 2003). Although the Dewagiriya

157

community felt that outside assistance was lacking, all participants felt that the Wildlife Department

should be responsible for keeping elephants away from their village (researcher pers. comm). Most

people felt that they lacked the financial or physical capacity to deter elephants from their homes,

and, for farmers who struggled with their fences, introducing a technique requiring the development

of new skills (beekeeping) may not have generated a sense of ‘empowerment’ and instead validated

the challenging nature of HEC mitigation.

4.6 Conclusion

Across all elephant range countries, numerous communities struggle to cope with crop-raiding

elephants, and the challenge of identifying feasible and effective elephant deterrents is real. Since it

was first developed and trialled in Kenya, beehive fencing has attracted interest as a practical eco-

deterrent that can be implemented at the community level. Despite obvious differences in the biology

and behaviour of the African and Asian elephant and honey bee species, interest in using beehive

fencing in Asia has been strong. Our trial provided the first long-term opportunity to evaluate the

practical application of beehive fencing as a deterrent for Asian elephants, incorporating not only

deterrent effect but also beekeeping challenges, cost effectiveness, and farmer motivation into future

recommendations.

Despite promising signs of being a relatively effective deterrent with a > 50% success rate at keeping

elephants from entering home gardens, it is unlikely that beehive fences could contribute

meaningfully to mitigating HEC at a community level in Dewagiriya Village. The difficulties in attracting

natural colonies, poor honey production and apparent lack of food resources for bees were consistent

over the study period, suggesting the saturation rate for supporting viable colonies in Dewagiriya

Village is low. Economically, farmers were unable to meet even small expenses such as replacing

damaged posts or hives, and financial support would be required for both initial set-up and ongoing

maintenance. Finally, although farmers seemed to enjoy participating in the project, motivation or

capacity to independently look after their fences was low, and it is unlikely that most of the existing

fences will continue to be used following this trial. Given the promising signs of deterrent success, it

would be worth exploring the potential of beehive fencing as an Asian elephant deterrent tool further.

However, future trials would benefit from careful selection of HEC location more amenable to

beekeeping.

158

4.7 References





Bandara, R & Tisdell, C 2002, 'Asian elephants as agricultural pests: economics of control and compensation in Sri Lanka', Natural Resources Journal, vol. 42(3), pp. 491-519.

Baishya, HK, Dey, S, Sarmah, A, Sharma, A, Gogoi, S, Aziz, T, Ghose, D & Williams, AC 2012, 'Use of chilli

fences to deter Asian elephants – a pilot study', Gajah, vol. 36, pp. 11 – 13.

Barnes, RFW, Danquah, E, Hema, EM, Dubiure, UF, Manford, M, Nandjui, A & Boafo, Y 2015,

'Retrospective versus prospective designs for studies of crop-raiding by elephants in Kakum, Ghana',


Barua, M, Bhagwat, SA & Jadhav, S 2013, 'The hidden dimensions of human-wildlife conflict: health

impacts, opportunity and transaction costs', Biological Conservation, vol. 157, pp. 309 – 316.







Chang’a, A, de Souza, N, Muya, J, Keyyu, J, Mwakatobe, A, Malugu, L, Ndossi, HP, Konuche, J, Omondi,

R, Mpinge, A, Hahn, N, Palminteri, S and Olson, D 2016, 'Scaling-up the use of chilli fences for reducing

human-elephant conflict across landscapes in Tanzania', Tropical Conservation Science, vol. 9, no. 2,

pp. 921 – 930.

159

Chiyo, PI, Cochrane, EP, Naughton, L & Basuta, GI 2005, 'Temporal patterns of crop raiding by

elephants: a response to changes in forage quality or crop availability?', African Journal of Ecology,

vol. 43, pp. 48 – 55.


9, no. 4, pp. 261 – 270.

Cook, RM, Witkowski, ETF, Helm, CV, Henley, MD & Parrini, F 2017, 'Recent exposure to African

elephants after a century of exclusion: rapid accumulation of marula tree impact and mortality, and

poor regeneration', Forest Ecology and Management, vol. 401, pp. 107 – 116.






doi: 10.1111/j.1755-263X.2011.00182.x

de Silva, M 1998, 'Status and conservation of the elephant (Elephas maximus) and the alleviation of

man-elephant conflict in Sri Lanka', Gajah, vol. 19, pp. 1 – 68.

Dickman, AJ 2010, 'Complexities of conflict: the importance of considering social factors for effectively

resolving human-wildlife conflict', Animal Conservation, vol. 13, pp. 458 - 466.

Dublin, HT & Hoare, RE 2004, 'Searching for solutions: the evolution of an integrated approach to

understanding and mitigating human-elephant conflict in Africa', Human Dimensions of Wildlife, vol.

9, no. 4, pp. 271 – 278.

Dutta, A 2013, ‘Bees, acrid smoke from chillis keep elephants away and reduce conflict’, The Times of

India, June 24, 2013.

160

Fernando, P, Leimgruber, P, Prasad, T & Pastorini, J 2012, ‘Problem-elephant translocation:

translocating the problem and the elephant?’, PLoS ONE, vol. 7, no. 12, e50917,

https://doi.org/10.1371/journal.pone.0050917.



Fernando, P, Kumar, MA, Williams, AC, Wikramanayake, E, Aziz, T & Singh, SM 2008, Review of human-

elephant conflict mitigation measures practiced in South Asia, AREAS Technical Support Document,

World Wide Fund for Nature.







Graham, MD & Ochieng, T 2008, 'Uptake and performance of farm-based measures for reducing crop-

raiding by elephants Loxodonta Africana among smallholder farms in Laikipia District, Kenya', Oryx,

vol. 42, pp. 76 – 82.

Gross, EM, Drouet-Hoguet, N, Subedi, N & Gross, J 2017, 'The potential of medicinal and aromatic

plants (MAPs) to reduce crop damages by Asian elephants (Elephas maximus)', Crop Protection, vol.

100, pp. 29 – 37.

Gross, EM, Mc Robb, R and Gross, J 2016, 'Cultivating alternative crops reduces crop losses due to

African elephants' Journal of Pest Science, vol. 89, no. 2, pp. 495 – 506.



161

Gunaryadi, D, Sugiyo, Hedges, S 2017, 'Community-based human-elephant conflict mitigation: the

value of an evidence-based approach in promoting the uptake of effective methods', PLoS ONE, vol.

12, no. 5, e0173742.

Hartter, J, Goldman, A & Southworth, J 2011, 'Responses by households to resource scarcity and

human-wildlife conflict: issues of fortress conservation and the surrounding agricultural landscape',

Journal for Nature Conservation, vol. 19, pp. 79 – 86.



Hedges, S & Gunaryadi, D 2009, 'Reducing human-elephant conflict: do chillies help deter elephants

from entering crop fields?', Oryx, vol. 44, no. 1, pp. 139 – 146.

Hoare, R 2015, ‘Lessons from 20 years of human-elephant conflict mitigation in Africa’, Human


Hoare, R 2000, 'African elephants and humans in conflict: the outlook for co-existence', Oryx, vol. 34,

pp. 34 – 38.

Hoffmeier-Karimi, RR & Schulte, BA 2014, 'Assessing perceived and documented crop damage in a

Tanzanian village impacted by human-elephant conflict (HEC)', Pachyderm, vol. 56, pp. 51 – 60.

Jasmine, B, Ghose, D and Das, SK 2015, 'An attitude assessment of human-elephant conflict in a critical

wildlife corridor within the Terai Arc landscape, India', Journal of Threatened Taxa, vol. 7, no. 2, pp.

6843 – 6852.

Karidozo, M & Osborn, FV 2015, 'Community based conflict mitigation trials: results of field tests of

chilli as an elephant deterrent', Biodiversity and Endangered Species, vol. 3 no. 1,

http://dx.doi.org/10.4172/2332-2543.1000144.

King, L, Pardo, M, Weerathunga, S, Kumara, TV, Jayasena, N, Soltis, J & de Silva, S 2018, ‘Wild Sri Lankan

elephants retreat from the sound of disturbed Asian honey bees’, Current Biology, vol. 28, pp 64 – 65.

162




King, LE 2014, Beehive fence construction manual: a step by step guide to building a protective beehive

fence to deter crop-raiding elephants from farm land, Elephants and Bees Project, Save the Elephants,

Nairobi.

King, LE 2010, The interaction between the African elephant (Loxodonta africana africana) and the

African honey bee (Apis mellifera scutellate) and its potential application as an elephant deterrent,

PhD Thesis, University of Oxford, UK.

King, LE, Soltis, J, Douglas-Hamilton, I, Savage, A & Vollrath, F 2010, 'Bee threat elicits alarm call in

African elephants', PLoS ONE, vol. 5, no. 4, e10346. doi:10.1371/journal.pone0010346.

King, LE, Lawrence, A, Douglas-Hamilton, I & Vollrath, F 2009, ‘Beehive fence deters crop-raiding

elephants’, African Journal of Ecology, vol. 47, pp. 131 – 137.

King, LE, Douglas-Hamilton, I & Vollrath, F 2007, 'African elephants run from the sounds of disturbed

bees', Current Biology, vol. 17, no. 19, R832.

Kioko, J, Muruthi, P, Omondi, P & Chiyo, PI 2008, 'The performance of electric fences as elephant

barriers in Amboseli, Kenya', South African Journal of Wildlife Research, vol. 38(1), pp. 52-58.

Koeniger, N & Koeniger, G 1994, Honeybees of Asia, IN: Ed (RWK Punchihewa), Beekeeping for honey

production in Sri Lanka, Sri Lanka Department of Agriculture, Peradeniya, Sri Lanka.

Mabelunga, T, Kumar, DAV, Gayathri, A & Krishnan, A 2016, 'Influence of elephant-human interactions

on agrarian communities in the Bengaluru-Bannerghatta Landscape – a perspective study', Gajah, vol.

45, pp. 28 – 32.

Madden, F 2004, 'Creating coexistence between humans and wildlife: global perspectives on local

efforts to address human-wildlife conflict', Human Dimensions of Wildlife, vol. 9, no. 4, pp. 247 – 257.

163

Manoj, EM 2012, ‘Jumbos get a sting of defeat’, The Hindu Newspaper, February 25, 2012.

Nath, NK, Lahkar, BP, Dutta, SK and Das, JP 2015, 'Human-elephant conflict around Manas National

Park, India: local people's attitudes, expectations and perceptions', Gajah, vol. 42, pp. 15 – 21.

Nair, RP and Jayson, EA 2016, 'Effectiveness of beehive fences to deter crop raiding elephants in

Kerala, India', International Research Journal of Natural and Applied Sciences, vol. 3, no. 1, pp. 14 – 19.

Nelson, A, Bidwell, P and Sillero-Zubiri 2003, 'A review of human-elephant conflict mitigation

strategies', People and Wildlife Initiative, Wildlife Conservation Research Unit, Oxford University.

Nsonsi, F, Heymans, J, Diamouangana, J, Mavinga, FB and Breuer, T 2018, 'Perceived human-elephant

conflict and its impact for elephant conservation in northern Congo', African Journal of Ecology, vol.

56, pp. 208 – 215.

Ogra, MV 2008, 'Human-wildlife conflict and gender in protected area borderlands: a case study of

costs, perceptions and vulnerabilities from Uttarakhand (Uttaranchal), India', Geoforum, vol. 39, pp.

1408 – 1422.



10.1017/90030605302000000.



Pozo, RA, Coulson, T, McCulloch, G, Strongza, A & Songhurst, A 2017, 'Chilli-briquettes modify the

temporal behaviours of elephants, but not their numbers', Oryx, doi: 10.1017/S0030605317001235.



pp. 4 – 11.

164

Punchihewa, RWK 1994, Beekeeping for honey production in Sri Lanka, Sri Lanka Department of

Agriculture, Peradeniya, Sri Lanka.

Ranasinghe, D, Ratnapala, S, Seneviratne, L, Schwammer, H, Schwammer, G & Perera, BH 2010a, Saga

of the elephant in Sri Lanka: extracts from Loris Volume 1 (1936) to Volume 10 (1964), YN Print

Graphics, Nedimala, Sri Lanka.

Ranasinghe, D, Ratnapala, S, Seneviratne, L, Schwammer, H, Schwammer, G & Perera, BH 2010b, Saga

of the elephant in Sri Lanka: extracts from Loris Volume 11 (1965) to Volume 20 (1993), YN Print

Graphics, Nedimala, Sri Lanka.

Santiapillai, C, Wijeyamohan, S, Bandara, G, Athurupana, R, Dissanayake, N & Read, B 2010, 'An

assessment of the human-elephant conflict in Sri Lanka', Ceylon Journal of Biological Sciences, vol. 39,

no. 1, pp. 21 – 33.



Santiapillai, C 1996, 'Mitigation of human-elephant conflicts in Sri Lanka', Gajah, vol. 15, pp. 1 – 7.

Scheijen, CPJ, Richards, SA, Smit, J, Jones, T & Nowak, K 2018, 'Efficacy of beehive fences as barriers

to African elephants: a case study in Tanzania', Oryx, doi: 10.1017/S0030605317001727.

Shri, KP, Gunasekaran, C, Salahudeen, M, Asaikkutti, A & Deepa, AA 2013, ‘Assessment of response of

elephants to beehives placed in a paddy field’, Gajah, vol. 38, pp. 42 – 43.



Sitati, NW, Walpole, MJ, Smith, RJ & Leader-Williams, N 2003, 'Predicting spatial aspects of human-

elephant conflict', Journal of Applied Ecology, vol. 40, pp. 667 – 677.

Sukumar, R 2006, 'A brief review of the status, distribution and biology of wild Asian elephants Elephas

maximus', International Zoo Yearbook, vol. 40, pp. 1 – 8.

165

Sukumar, R 1990, 'Ecology of the Asian elephant in southern India. II. Feeding habits and crop raiding

patterns', Journal of Tropical Ecology, vol. 6, pp. 33 – 53.



Thirgood, S, Woodroffe, R and Rabinowitz, A 2005, 'The impact of human-wildlife conflict on human

lives and livelihoods', IN: (eds: Woodroffe R, Thirgood S and Rabinowitz A) People and wildlife: conflict

or coexistence? Cambridge, Cambridge University Press, pp. 13 – 26.

van de Water, A and Matteson, K 2018, 'Human-elephant conflict in western Thailand: socio-economic

drivers and potential mitigation strategies', PLoS ONE, vol. 13, no. 6, e0194736.

van Eden, M, Ellis, E and Bruyere, BL 2016, 'The influence of human-elephant conflict on electric fence

management and perception among different rural communities in Laikipia County, Kenya', Human


Vollrath, F & Douglas-Hamilton, I 2002, 'African bees to control African elephants',

Naturwissenschaften, vol. 89, pp. 508 – 511.





243 – 251.






pp. 164-169.

166

Chapter 5: Discussion

167

5.1 Introduction

The research presented here constitutes a multi-faceted investigation, where we attempt to better

understand the behaviour of a local elephant population in north-central Sri Lanka, the patterns of

crop-raiding, and to work closely with a local community heavily impacted by HEC to evaluate a newly

introduced elephant deterrent tool. To be more specific, we investigated the general behaviour

patterns of elephants utilising habitat representing two different levels of anthropogenic disturbance

or ‘risk’ to elephants (Wasgamuwa National Park: low risk; Weheragala Tank: medium risk), in the

presence and absence of an immediate disturbance stimuli. We also monitored crop-raiding activity

over a three-year period to produce a comprehensive profile of the crop-raiding patterns and

predictors occurring in Dewagiriya Village, a small community heavily impacted by HEC. Finally, in

collaboration with farmers from the Dewagiriya community, we established, monitored and evaluated

the practical feasibility of an elephant deterrent new to the area: the use of beehive fences to protect

home gardens.

In this chapter, I begin with a brief summary of research findings. I then discuss methodological and

project challenges. Finally, I suggest future research directions that I believe would help further our

understanding of complex HEC situations and complement the development of site-specific mitigation

planning to benefit both elephants and humans.

5.2 Summary of results

By scoring the duration of general behavioural activities (feeding, locomotion, standing still) and

frequency of specific ‘reactive’ behaviours (e.g. head-shaking, dusting, backing away) from video

footage obtained during direct field observations, we were able to compare the behaviour of male

and female elephants inhabiting areas within Wasgamuwa National Park (a low-risk zone) and at

Weheragala Tank (a medium-risk zone). We were also able to identify whether responses to

immediate disturbances in the environment (e.g. approach of vehicles, pedestrians or dogs) were

consistent across the two risk zones.

We found no evidence to suggest that general behaviour patterns of elephants differed according to

the perceived level of anthropogenic risk (medium or low) represented by their location. Not

168

surprisingly, elephants spent the majority of their time feeding, whether they were observed in

Wasgamuwa National Park or at Weheragala Tank. However, the presence of an immediate

anthropogenic risk or disturbance, did cause elephants to alter their behaviour, with this effect

stronger at Weheragala Tank. At both locations, the frequency of ‘reactive behaviours’ (typically

smelling or assuming the ‘vigilance’ posture) increased. However, at Weheragala Tank elephants of

both sexes spent less time feeding during or following a disturbance event. This is in line with a study

in India where elephants also decreased feeding time as levels of human disturbance increased

(Srinivassaiah et al. 2012). In the presence of a disturbance, female elephants also increased the time

they spent in locomotion, often displaying avoidance behaviour as they interrupted feeding to retreat

back to the relative safety of forest cover.

Farmer-collected data was used to document sightings of elephants within 50 m of their property,

including date, time, number of elephants, sex of elephants, and damage incurred. Crop-raiding

patterns and characteristics were measured from these data. Our results show that Dewagiriya Village

experiences exceptionally high levels of crop-raiding, especially when compared to other locations

within both Africa and Asia, based on a review of available literature (examples: Das et al. 2012; Davies

et al. 2011; Ekanayaka et al. 2011; Campos-Arceiz et al. 2009; Barnes et al. 2007; Chen et al. 2005).

Several characteristics of crop-raiding in Dewagiriya were typical of most crop-raiding situations: In

particular, our study showed that almost all crop-raiding occurred at night and evidence strongly

suggested that male elephants were the predominant crop-raiders, acting either individually or in

small bachelor groups (for comparison see examples: Thaufeek et al. 2014; Das et al. 2014; Graham et

al. 2010; Campos-Arceiz et al. 2009; Bandara and Tisdell 2003). Elephants raided a wide variety of

crops, with rice, bananas and coconuts the most frequently targeted, presumably at least in part

because they were the most widely available (for comparison see examples: Haturusinghe and

Weerakoon 2012; Ekanayaka et al. 2011; Campos-Arceiz et al. 2009; Wikramanayake et al. 2004).

Dewagiriya differs from most other locations (examples: Gubbi 2012; Webber et al. 2011; Campose-

Arceiz et al. 2009; Fernando et al. 2005), in that no clear seasonal patterns were identified. Elephants

raided year-round and no overt influence of season or rainfall was apparent, aside from a reduction

in crop-raiding events when it had rained during that 24-hour period. Clear within-site variations in

crop-raiding were evident, with homes and fields located furthest from the road and closest to the

water tanks and forested habitat, bearing the highest burden of elephant events and crop-raids.

169

During our three-year beehive fence trial we worked closely with the local community to decide on

experimental participants, design fence layouts, build and monitor the fences, and record data on

elephant events occurring within 50 m of home garden boundaries for experimental (beehive fence)

and control (no beehive fence) properties. Results provided evidence that beehive fences were

effective at reducing elephant incursions into home gardens relative to control home gardens.

However, there was still a 43% rate of elephants breaching fence perimeters on approach. Still, the

deterrence effect strengthened with increased hive occupations, indicating that the presence of bees

was influential in preventing elephants from entering the fences. Had we been able to reach our 50%

hive occupation target on more fences, less breakthroughs may have occurred. That said, beekeeping

was challenging and attracting wild colonies into our fence hives was largely unsuccessful. Colony

health was often weak and as a result honey production was low and no economic benefits were seen.

Farmer motivation fluctuated and only four experimental participants planned to continue using their

fences after the trial ended.

5.3 Methodological challenges

5.3.1 Behavioural assessments

Studying the behaviour of wild animals in a natural setting can be fraught with challenges, the most

obvious being an inability to account for all confounding variables. Numerous factors, ranging from

subtle variations in environmental conditions (Griffin 2007) to the personalities of individuals within

the group (Hollander et al. 2008; Freeman et al. 2007) can influence behaviour. In an uncontrolled

environment, such as our study sites of Wasgamuwa National Park and Weheragala Tank, we were

unable to select specific individuals for observation, resulting in unintentional bias towards inclusion

of individuals who were ‘bold’ enough to enter the observation space in the presence of our research

vehicle. It is entirely possible that more elephants from the same group or family as those observed

were present during observations but did not leave the relative safety of forest cover. Another

potential issue is that we had little or no knowledge of events that happened previously that day (or

earlier) that may have influenced behaviour outside of our parameters of interest. Another constraint

was our reliance on opportunistic disturbance, and we had no control over what particular disturbance

occurred or which elephants were present at the time. This limited our sample size of study subjects

and the contexts in which they could be observed. Still captive studies cannot be expected to

accurately replicate the ‘wild’ environment, and natural studies are important. Comparing responses

170

of captive elephants during exposure to anthropogenically induced disturbances would have its own

ethical considerations, and relevance to a ‘real life’ setting would be limited. We dealt with these

constraints by randomly selecting study subjects and disturbance events and acknowledge that

conclusions rely on a small degree of speculation.

In this study we made the decision to include only a ‘low’ and ‘medium’ risk zone (Wasgamuwa

National Park and Weheragala Tank respectively). Including a ‘high’ risk zone would have added an

extra layer of complexity to our study, allowing us to compare elephant behaviour at two ‘extremes’

of anthropogenic disturbance, whilst still investigating potential impacts of ‘medium’ level

disturbance. Dewagiriya Village, where the crop-raiding profile and beehive fence trial were

conducted, seemed an obvious candidate for a ‘high’ risk zone. Elephants were known to inhabit the

forest on the periphery of Dewagiriya Village during the day and move towards the village to drink

from the water tanks and crop-raid in the evenings. However, the thick forest habitat was accessible

only on foot and no clearings suitable for video recording were available. Observing elephants at night

would have reduced comparability across risk zones as all other observations were conducted

between 3 pm and 7 pm. If differences in behaviour had been observed, it would have been difficult

to tease apart the influence of time on behaviour versus the influence of heightened anthropogenic

risk. Furthermore, community members did not want our team conducting observations in Dewagiriya

Village at night, ostensibly (and reasonably) for our own safety but I speculate that certain farmers

within the community did not want outsiders observing their personal interactions with elephants as

illegal methods are sometimes used to chase them out of the village.

I would encourage researchers at other HEC locations to investigate the behaviour of elephants in

‘medium’ risk zones. To take Sri Lanka as an example, approximately 65% of suitable elephant habitat

is located outside of protected boundaries (Santiapillai et al. 2006), and most elephants do venture

outside of PA’s to utilise other suitable habitat (Weerakoon et al. 2004). Elephants tend to visit villages

(or high risk zones) only to crop raid which means a lot of time is spent in areas akin to our ‘medium’

risk zone. In Dewagiriya Village, humans and elephants both utilise water and forest resources, and

use the paths and roads to travel from village to village. Knowledge of the general behaviour patterns

of elephants, and their responses to human presence or disturbance (whether on foot or in vehicles)

can help facilitate safe interactions between the two species, outside of the intense conflict situations

occurring nightly when elephants come to feed on crops.

171

5.3.2 Crop-raiding activity

To identify crop-raiding activity in Dewagiriya Village, we relied heavily on information from farmers.

We tried to make this as systematic as possible, but most families did not fill out their data sheets,

and we found it more accurate to visit bi-monthly and transcribe information on-the-spot. Calendars

were distributed, so people could at least mark the date of elephant events to ensure none were

forgotten but only some families used this. Still, an elephant event was a memorable occasion, despite

the frequency with which they occurred, and generally people were keen to provide reports. To the

best of our knowledge, elephant events were reported with accuracy, and reports from neighbouring

households typically corroborated. Even allowing for some degree of exaggeration, the frequency of

elephant events in Dewagiriya Village is undeniably high, and a major social concern for the

community.

Admittedly, our study had a large focus on frequency of elephant events and quantifying precise

measurements of crop damage was outside our scope. Further research quantifying crop damage

relative to overall size of crop plots and the stage of growth when damaged/impacts on crop yields

would facilitate economic impact assessments of crop-raids. We simply did not have enough time or

manpower to go into this detail. It was also obvious that even small crop losses were significant to

individual households.

5.3.3 Beehive fence trials

Taking a concept designed for another species of bees and another species of elephants (King et al.

2011) was always going to be challenging but given the success shown in Africa and that beekeeping

can produce additional benefits (e.g. honey production), we felt it was an idea well worth attempting.

However, although we found indicators of success at preventing elephants from entering home

gardens, the deterrent rate was not as strong as hoped. This combined with the challenges of

beekeeping, fence costs, and low farmer motivation perhaps suggests that the observed positives do

not outweigh the challenges, at least for Dewagiriya Village.

To rigorously test the efficacy of beehive fences as an elephant deterrent, we would ideally leave

occupied fences as the only line of defence. However, in the real-life context of Dewagiriya Village it

would not have been fair to ask farmers to eliminate use of all traditional methods. Elephants can

172

easily destroy numerous crops or plants in a single visit, plus allowing elephants to approach when

family members are inside their homes is a huge risk to human safety (Fernando et al. 2005). Instead

we asked farmers to list other deterrents used and found the same methods used at both

experimental and control households. This typically involved shouting, throwing firecrackers or

thunder-flashes, and shining bright flashlights in the direction of elephants. However, we were unable

to measure farmer vigilance and it’s possible that households who guarded more closely or had more

manpower were better able to defend their property regardless of whether they had a beehive fence,

or how many occupied hives were present.

The only way to potentially combat this methodological challenge, would be to plant a series of crop

plots outside of the village using project funds, and surround these with beehive fences. This would

negate the risk to farmer livelihoods and personal safety. However, finding an unplanted and

unforested area large enough to support multiple beehive fences (and getting permission to use it)

would be difficult. Furthermore, because elephants are known to use a combination of olfactory,

audio and visual cues to distinguish threats (Bates et al. 2007; Thuppil and Coss 2013), such a

fabricated environment would fail to fully represent the in-village crop-raiding scenario where human

influences such as lights, sounds and smell are always present. Perhaps the best way to answer the

question of whether beehive fencing is a useful Asian elephant deterrent is in the exact context where

the complexities and variables involved in mitigation conflict exist.

It should also be noted that in the lifespan of either a human or elephant, 3 years is a relatively short

time. Although we consider this a sufficient time span to provide indicators of the likelihood of beehive

fence success, a much longer-term study is recommended to gain a true sense of its worthiness in an

overall human-elephant conflict toolkit. Over time elephants could habituate to the presence of bees

and find more ways to breach the deterrent, as they have been known to do with other mitigation

techniques (examples: Fernando et al. 2011; Osborn and Parker 2003). Alternatively, elephants could

learn more about the potential negative consequences of disturbing beehives and be less likely to

break-through. Farmer willingness to participate could also change over time. If beekeeping skills

improved, and honey harvests were higher, this could be a major incentive to maintain fences in good

condition.

173

Although we did our best to ensure the community were empowered with the skills necessary to look

after their fences and bee colonies, after three years only two households could manage their fences

and beekeeping independently. Two households did not have the time and were not confident with

beekeeping, and therefore withdrew from the trial. The other households fell somewhere in the

middle, needing some active support and reminding of maintenance tasks but also capable of doing

most things alone. Another limitation to farmers continuing with the deterrent is that no households

were in a financial position to afford to replace damaged posts or beehives, or to introduce new

colonies if their colonies absconded. Overall, only four households planned to continue using their

beehive fences post-trial. We had hoped three years would be a sufficient period to properly establish

the beehive fence project and transfer management to the community, but this did not eventuate.

Efforts were made to connect with local organisations who could continue to provide small-scale

practical and financial support in fence maintenance and beekeeping, following our three-year trial

period. However, despite expressed enthusiasm and training provided, to the best of our knowledge

this has not been implemented.

5.4 Closing statements

This study has contributed to the general body of knowledge on elephant behaviour in an

anthropogenically influenced context, and specifically on patterns of Asian elephant crop-raiding in a

high HEC zone, and the potential of beehive fencing as an Asian elephant deterrent. We show that at

our particular study site, elephants do make behavioural choices based on the presence of

anthropogenic disturbance in their immediate environment; that crop-raiding in Dewagiriya Village is

characterised by similar sex-biases and nocturnal timing as most locations but lacks clear seasonal

patterns; and that beehive fencing shows some positive indicators of deterrent success however is

limited by beekeeping challenges, cost-effectiveness, and farmer motivation.

There is no doubt that the Dewagiriya community is struggling to cope with HEC and given personal

observations from other nearby villages and the available HEC literature spanning elephant range

countries, it is fair to say that too many communities are struggling. The conversations about crop-

raiding and mitigation often seem to go in circles, although it is commonly agreed that genuine

collaboration between stakeholders (government authorities, NGO’s, wildlife managers,

conservationists, and local communities) is necessary to enact meaningful change. However,

174

translating this theoretical realisation into on-ground action is difficult. Still, we need to continue to

do the best we can to help individuals and communities, and to generate information that might be

of use to others. For elephant conservation to succeed, cooperation rather than pushing individual

agendas, and free sharing of research findings (positive or negative results) is one of the most powerful

weapons we have. Finally, we must remember that the well-being and survival of people who share

their land with elephants is equally as important as the charismatic species we are trying so

desperately to protect.

175

5.5 References





doi:10.1016/j.cub.2007.09.060.




Chen, Y, Marino, J, Chen, Y, Tao, Q, Sullivan, CD, Shi, K & MacDonald, DW 2016, 'Predicting hotspots

of human-elephant conflict to inform mitigation strategies in Xishuangbanna, Southwest China', PLoS

ONE, vol. 11(9): e0162035.doi:10.1371/journal.pone.0162035.






doi: 10.1111/j.1755-263X.2011.00182.x



pp. 20 -25.



176




Freeman, HD, Brosnan, SF, Hopper, LM, Lambeth, SP, Schapiro, SJ & Gosling, SD 2013, 'Developing a

comprehensive and comparative questionnaire for measuring personality in chimpanzees using a

simultaneous top-down/bottom-up design', American Journal of Primatology, vol. 75, pp. 1042 –

1053.




Griffin, SC, Valois, T, Taper, ML & Mills, LS 2007, 'Effects of tourists on behaviour and demography of

Olympic marmots', Conservation Biology, vol. 21, no. 4, pp. 1070-1081.





Hollander, FA, van Overveld, T, Tokka, I & Matthysen, E 2008, ‘Personality and nest defence in the

great tit (Parsus major)’, Ethology, vol. 114, pp. 405 – 412.




King, LE, Douglas-Hamilton, I & Vollrath, F 2011, 'Beehive fences as effective deterrents for crop-

raiding elephants: field trials in northern Kenya', African Journal of Ecology, doi: 10.1111/j.1365-

2028.2011.01275.x.

177



10.1017/90030605302000000.



Srinivasaiah, NM, Anand, VD, Vaidyanathan, S & Sinha, A 2012, 'Usual populations, unusual

individuals: insights into the behaviour and management of Asian elephants in fragmented

landscapes', PLoS ONE, vol. 7, no. 8, e42571. doi: 10.1371/journal.pone.0042571.

Thaufeek, UL, Padmalal, UKGK and Fernando, P 2014, 'Land use and human-elephant conflict in the

Sigiriya Sanctuary, Sri Lanka', Gajah, vol. 40, pp. 26 – 30.

Thuppil, V and Coss RG 2013, 'Wild Asian elephants distinguish aggressive tiger and leopard growls

according to perceived danger', Biology Letters, Vol. 9, 20130518.




243 – 251.











pp. 164-169.

http://dx.doi.org/10.1098/rsbl.2013.0518

Can beehive fences help mitigate human-elephant confli

Documents

Transcript of Can beehive fences help mitigate human-elephant confli